open-consul/agent/agent_endpoint_test.go

7143 lines
194 KiB
Go
Raw Normal View History

package agent
import (
"bytes"
"context"
"crypto/tls"
2018-03-21 17:55:39 +00:00
"crypto/x509"
"encoding/json"
"fmt"
2016-11-16 21:45:26 +00:00
"io"
"io/ioutil"
"net/http"
"net/http/httptest"
"net/url"
"os"
"strconv"
"strings"
"testing"
"time"
2015-05-11 23:48:10 +00:00
2021-07-15 23:04:24 +00:00
"github.com/armon/go-metrics"
"github.com/hashicorp/go-hclog"
"github.com/hashicorp/go-uuid"
"github.com/hashicorp/serf/serf"
"github.com/mitchellh/hashstructure"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"golang.org/x/time/rate"
"github.com/hashicorp/consul/acl"
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
"github.com/hashicorp/consul/agent/config"
2018-03-21 17:20:35 +00:00
"github.com/hashicorp/consul/agent/connect"
"github.com/hashicorp/consul/agent/connect/ca"
"github.com/hashicorp/consul/agent/consul"
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
"github.com/hashicorp/consul/agent/debug"
"github.com/hashicorp/consul/agent/local"
"github.com/hashicorp/consul/agent/structs"
"github.com/hashicorp/consul/agent/token"
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
tokenStore "github.com/hashicorp/consul/agent/token"
"github.com/hashicorp/consul/agent/xds/proxysupport"
"github.com/hashicorp/consul/api"
"github.com/hashicorp/consul/sdk/testutil"
"github.com/hashicorp/consul/sdk/testutil/retry"
"github.com/hashicorp/consul/testrpc"
"github.com/hashicorp/consul/types"
)
func createACLTokenWithAgentReadPolicy(t *testing.T, srv *HTTPHandlers) string {
policyReq := &structs.ACLPolicy{
Name: "agent-read",
Rules: `agent_prefix "" { policy = "read" }`,
}
req, _ := http.NewRequest("PUT", "/v1/acl/policy?token=root", jsonReader(policyReq))
resp := httptest.NewRecorder()
srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
tokenReq := &structs.ACLToken{
Description: "agent-read-token-for-test",
Policies: []structs.ACLTokenPolicyLink{{Name: "agent-read"}},
}
req, _ = http.NewRequest("PUT", "/v1/acl/token?token=root", jsonReader(tokenReq))
resp = httptest.NewRecorder()
srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
svcToken := &structs.ACLToken{}
dec := json.NewDecoder(resp.Body)
err := dec.Decode(svcToken)
require.NoError(t, err)
return svcToken.SecretID
}
func TestAgent_Services(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
srv1 := &structs.NodeService{
ID: "mysql",
Service: "mysql",
Tags: []string{"primary"},
Meta: map[string]string{
"foo": "bar",
},
Port: 5000,
}
require.NoError(t, a.State.AddService(srv1, ""))
req, _ := http.NewRequest("GET", "/v1/agent/services", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
decoder := json.NewDecoder(resp.Body)
var val map[string]*api.AgentService
err := decoder.Decode(&val)
require.NoError(t, err)
assert.Lenf(t, val, 1, "bad services: %v", val)
assert.Equal(t, 5000, val["mysql"].Port)
assert.Equal(t, srv1.Meta, val["mysql"].Meta)
}
func TestAgent_ServicesFiltered(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
srv1 := &structs.NodeService{
ID: "mysql",
Service: "mysql",
Tags: []string{"primary"},
Meta: map[string]string{
"foo": "bar",
},
Port: 5000,
}
require.NoError(t, a.State.AddService(srv1, ""))
// Add another service
srv2 := &structs.NodeService{
ID: "redis",
Service: "redis",
Tags: []string{"kv"},
Meta: map[string]string{
"foo": "bar",
},
Port: 1234,
}
require.NoError(t, a.State.AddService(srv2, ""))
req, _ := http.NewRequest("GET", "/v1/agent/services?filter="+url.QueryEscape("foo in Meta"), nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
decoder := json.NewDecoder(resp.Body)
var val map[string]*api.AgentService
err := decoder.Decode(&val)
require.NoError(t, err)
require.Len(t, val, 2)
req, _ = http.NewRequest("GET", "/v1/agent/services?filter="+url.QueryEscape("kv in Tags"), nil)
resp = httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
decoder = json.NewDecoder(resp.Body)
val = make(map[string]*api.AgentService)
err = decoder.Decode(&val)
require.NoError(t, err)
require.Len(t, val, 1)
}
2018-03-12 20:05:06 +00:00
// This tests that the agent services endpoint (/v1/agent/services) returns
// Connect proxies.
func TestAgent_Services_ExternalConnectProxy(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
srv1 := &structs.NodeService{
Add Proxy Upstreams to Service Definition (#4639) * Refactor Service Definition ProxyDestination. This includes: - Refactoring all internal structs used - Updated tests for both deprecated and new input for: - Agent Services endpoint response - Agent Service endpoint response - Agent Register endpoint - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Register - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Services endpoint response - Catalog Node endpoint response - Catalog Service endpoint response - Updated API tests for all of the above too (both deprecated and new forms of register) TODO: - config package changes for on-disk service definitions - proxy config endpoint - built-in proxy support for new fields * Agent proxy config endpoint updated with upstreams * Config file changes for upstreams. * Add upstream opaque config and update all tests to ensure it works everywhere. * Built in proxy working with new Upstreams config * Command fixes and deprecations * Fix key translation, upstream type defaults and a spate of other subtele bugs found with ned to end test scripts... TODO: tests still failing on one case that needs a fix. I think it's key translation for upstreams nested in Managed proxy struct. * Fix translated keys in API registration. ≈ * Fixes from docs - omit some empty undocumented fields in API - Bring back ServiceProxyDestination in Catalog responses to not break backwards compat - this was removed assuming it was only used internally. * Documentation updates for Upstreams in service definition * Fixes for tests broken by many refactors. * Enable travis on f-connect branch in this branch too. * Add consistent Deprecation comments to ProxyDestination uses * Update version number on deprecation notices, and correct upstream datacenter field with explanation in docs
2018-09-12 16:07:47 +00:00
Kind: structs.ServiceKindConnectProxy,
ID: "db-proxy",
Service: "db-proxy",
Port: 5000,
Proxy: structs.ConnectProxyConfig{
DestinationServiceName: "db",
Upstreams: structs.TestUpstreams(t),
},
}
a.State.AddService(srv1, "")
req, _ := http.NewRequest("GET", "/v1/agent/services", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
decoder := json.NewDecoder(resp.Body)
var val map[string]*api.AgentService
err := decoder.Decode(&val)
require.NoError(t, err)
assert.Len(t, val, 1)
2018-03-11 16:31:31 +00:00
actual := val["db-proxy"]
assert.Equal(t, api.ServiceKindConnectProxy, actual.Kind)
assert.Equal(t, srv1.Proxy.ToAPI(), actual.Proxy)
}
// Thie tests that a sidecar-registered service is returned as expected.
func TestAgent_Services_Sidecar(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
srv1 := &structs.NodeService{
Kind: structs.ServiceKindConnectProxy,
ID: "db-sidecar-proxy",
Service: "db-sidecar-proxy",
Port: 5000,
// Set this internal state that we expect sidecar registrations to have.
LocallyRegisteredAsSidecar: true,
Proxy: structs.ConnectProxyConfig{
DestinationServiceName: "db",
Upstreams: structs.TestUpstreams(t),
Mode: structs.ProxyModeTransparent,
TransparentProxy: structs.TransparentProxyConfig{
OutboundListenerPort: 10101,
},
},
}
a.State.AddService(srv1, "")
req, _ := http.NewRequest("GET", "/v1/agent/services", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
decoder := json.NewDecoder(resp.Body)
var val map[string]*api.AgentService
err := decoder.Decode(&val)
require.NoError(t, err)
assert.Len(t, val, 1)
actual := val["db-sidecar-proxy"]
require.NotNil(t, actual)
assert.Equal(t, api.ServiceKindConnectProxy, actual.Kind)
assert.Equal(t, srv1.Proxy.ToAPI(), actual.Proxy)
// Sanity check that LocalRegisteredAsSidecar is not in the output (assuming
2018-11-02 17:00:39 +00:00
// JSON encoding). Right now this is not the case because the services
// endpoint happens to use the api struct which doesn't include that field,
// but this test serves as a regression test incase we change the endpoint to
// return the internal struct later and accidentally expose some "internal"
// state.
assert.NotContains(t, resp.Body.String(), "LocallyRegisteredAsSidecar")
assert.NotContains(t, resp.Body.String(), "locally_registered_as_sidecar")
}
// This tests that a mesh gateway service is returned as expected.
func TestAgent_Services_MeshGateway(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
srv1 := &structs.NodeService{
Kind: structs.ServiceKindMeshGateway,
ID: "mg-dc1-01",
Service: "mg-dc1",
Port: 8443,
Proxy: structs.ConnectProxyConfig{
Config: map[string]interface{}{
"foo": "bar",
},
},
}
a.State.AddService(srv1, "")
req, _ := http.NewRequest("GET", "/v1/agent/services", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
var val map[string]*api.AgentService
err := dec.Decode(&val)
require.NoError(t, err)
require.Len(t, val, 1)
actual := val["mg-dc1-01"]
require.NotNil(t, actual)
require.Equal(t, api.ServiceKindMeshGateway, actual.Kind)
// Proxy.ToAPI() creates an empty Upstream list instead of keeping nil so do the same with actual.
if actual.Proxy.Upstreams == nil {
actual.Proxy.Upstreams = make([]api.Upstream, 0)
}
require.Equal(t, srv1.Proxy.ToAPI(), actual.Proxy)
}
// This tests that a terminating gateway service is returned as expected.
func TestAgent_Services_TerminatingGateway(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
srv1 := &structs.NodeService{
Kind: structs.ServiceKindTerminatingGateway,
ID: "tg-dc1-01",
Service: "tg-dc1",
Port: 8443,
Proxy: structs.ConnectProxyConfig{
Config: map[string]interface{}{
"foo": "bar",
},
},
}
require.NoError(t, a.State.AddService(srv1, ""))
req, _ := http.NewRequest("GET", "/v1/agent/services", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
var val map[string]*api.AgentService
err := dec.Decode(&val)
require.NoError(t, err)
require.Len(t, val, 1)
actual := val["tg-dc1-01"]
require.NotNil(t, actual)
require.Equal(t, api.ServiceKindTerminatingGateway, actual.Kind)
// Proxy.ToAPI() creates an empty Upstream list instead of keeping nil so do the same with actual.
if actual.Proxy.Upstreams == nil {
actual.Proxy.Upstreams = make([]api.Upstream, 0)
}
require.Equal(t, srv1.Proxy.ToAPI(), actual.Proxy)
}
func TestAgent_Services_ACLFilter(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
services := []*structs.NodeService{
{
ID: "web",
Service: "web",
Port: 5000,
},
{
ID: "api",
Service: "api",
Port: 6000,
},
}
for _, s := range services {
a.State.AddService(s, "")
}
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("GET", "/v1/agent/services", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
var val map[string]*api.AgentService
err := dec.Decode(&val)
if err != nil {
t.Fatalf("Err: %v", err)
}
if len(val) != 0 {
t.Fatalf("bad: %v", val)
}
require.Len(t, val, 0)
require.Empty(t, resp.Header().Get("X-Consul-Results-Filtered-By-ACLs"))
})
t.Run("limited token", func(t *testing.T) {
token := testCreateToken(t, a, `
service "web" {
policy = "read"
}
`)
req := httptest.NewRequest("GET", fmt.Sprintf("/v1/agent/services?token=%s", token), nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
var val map[string]*api.AgentService
if err := dec.Decode(&val); err != nil {
t.Fatalf("Err: %v", err)
}
require.Len(t, val, 1)
require.NotEmpty(t, resp.Header().Get("X-Consul-Results-Filtered-By-ACLs"))
})
t.Run("root token", func(t *testing.T) {
req, _ := http.NewRequest("GET", "/v1/agent/services?token=root", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
var val map[string]*api.AgentService
err := dec.Decode(&val)
if err != nil {
t.Fatalf("Err: %v", err)
}
require.Len(t, val, 2)
require.Empty(t, resp.Header().Get("X-Consul-Results-Filtered-By-ACLs"))
})
}
func TestAgent_Service(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig()+`
services {
name = "web"
port = 8181
tagged_addresses {
wan {
address = "198.18.0.1"
port = 1818
}
}
}
`)
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
proxy := structs.TestConnectProxyConfig(t)
proxy.DestinationServiceID = "web1"
// Define a valid local sidecar proxy service
sidecarProxy := &structs.ServiceDefinition{
Kind: structs.ServiceKindConnectProxy,
Name: "web-sidecar-proxy",
Check: structs.CheckType{
TCP: "127.0.0.1:8000",
Interval: 10 * time.Second,
},
Port: 8000,
Proxy: &proxy,
Weights: &structs.Weights{
Passing: 1,
Warning: 1,
},
EnterpriseMeta: *structs.DefaultEnterpriseMetaInDefaultPartition(),
}
// Define an updated version. Be careful to copy it.
updatedProxy := *sidecarProxy
updatedProxy.Port = 9999
// Mangle the proxy config/upstreams into the expected for with defaults and
// API struct types.
expectProxy := proxy
expectProxy.Upstreams =
structs.TestAddDefaultsToUpstreams(t, sidecarProxy.Proxy.Upstreams, *structs.DefaultEnterpriseMetaInDefaultPartition())
expectedResponse := &api.AgentService{
Kind: api.ServiceKindConnectProxy,
ID: "web-sidecar-proxy",
Service: "web-sidecar-proxy",
Port: 8000,
Proxy: expectProxy.ToAPI(),
Weights: api.AgentWeights{
Passing: 1,
Warning: 1,
},
Meta: map[string]string{},
Tags: []string{},
Datacenter: "dc1",
}
fillAgentServiceEnterpriseMeta(expectedResponse, structs.DefaultEnterpriseMetaInDefaultPartition())
hash1, err := hashstructure.Hash(expectedResponse, nil)
require.NoError(t, err, "failed to generate hash")
expectedResponse.ContentHash = fmt.Sprintf("%x", hash1)
// Copy and modify
updatedResponse := *expectedResponse
updatedResponse.Port = 9999
updatedResponse.ContentHash = "" // clear field before hashing
hash2, err := hashstructure.Hash(updatedResponse, nil)
require.NoError(t, err, "failed to generate hash")
updatedResponse.ContentHash = fmt.Sprintf("%x", hash2)
// Simple response for non-proxy service registered in TestAgent config
expectWebResponse := &api.AgentService{
ID: "web",
Service: "web",
Port: 8181,
Weights: api.AgentWeights{
Passing: 1,
Warning: 1,
},
TaggedAddresses: map[string]api.ServiceAddress{
"wan": {
Address: "198.18.0.1",
Port: 1818,
},
},
Meta: map[string]string{},
Tags: []string{},
Datacenter: "dc1",
}
fillAgentServiceEnterpriseMeta(expectWebResponse, structs.DefaultEnterpriseMetaInDefaultPartition())
hash3, err := hashstructure.Hash(expectWebResponse, nil)
require.NoError(t, err, "failed to generate hash")
expectWebResponse.ContentHash = fmt.Sprintf("%x", hash3)
tests := []struct {
name string
tokenRules string
url string
updateFunc func()
wantWait time.Duration
wantCode int
wantErr string
wantResp *api.AgentService
}{
{
name: "simple fetch - proxy",
url: "/v1/agent/service/web-sidecar-proxy",
wantCode: 200,
wantResp: expectedResponse,
},
{
name: "simple fetch - non-proxy",
url: "/v1/agent/service/web",
wantCode: 200,
wantResp: expectWebResponse,
},
{
name: "blocking fetch timeout, no change",
url: "/v1/agent/service/web-sidecar-proxy?hash=" + expectedResponse.ContentHash + "&wait=100ms",
wantWait: 100 * time.Millisecond,
wantCode: 200,
wantResp: expectedResponse,
},
{
name: "blocking fetch old hash should return immediately",
url: "/v1/agent/service/web-sidecar-proxy?hash=123456789abcd&wait=10m",
wantCode: 200,
wantResp: expectedResponse,
},
{
name: "blocking fetch returns change",
url: "/v1/agent/service/web-sidecar-proxy?hash=" + expectedResponse.ContentHash,
updateFunc: func() {
time.Sleep(100 * time.Millisecond)
// Re-register with new proxy config, make sure we copy the struct so we
// don't alter it and affect later test cases.
req, _ := http.NewRequest("PUT", "/v1/agent/service/register?token=root", jsonReader(updatedProxy))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, 200, resp.Code, "body: %s", resp.Body.String())
},
wantWait: 100 * time.Millisecond,
wantCode: 200,
wantResp: &updatedResponse,
},
{
// This test exercises a case that caused a busy loop to eat CPU for the
// entire duration of the blocking query. If a service gets re-registered
// wth same proxy config then the old proxy config chan is closed causing
// blocked watchset.Watch to return false indicating a change. But since
// the hash is the same when the blocking fn is re-called we should just
// keep blocking on the next iteration. The bug hit was that the WatchSet
// ws was not being reset in the loop and so when you try to `Watch` it
// the second time it just returns immediately making the blocking loop
// into a busy-poll!
//
// This test though doesn't catch that because busy poll still has the
// correct external behavior. I don't want to instrument the loop to
// assert it's not executing too fast here as I can't think of a clean way
// and the issue is fixed now so this test doesn't actually catch the
// error, but does provide an easy way to verify the behavior by hand:
// 1. Make this test fail e.g. change wantErr to true
// 2. Add a log.Println or similar into the blocking loop/function
// 3. See whether it's called just once or many times in a tight loop.
name: "blocking fetch interrupted with no change (same hash)",
url: "/v1/agent/service/web-sidecar-proxy?wait=200ms&hash=" + expectedResponse.ContentHash,
updateFunc: func() {
time.Sleep(100 * time.Millisecond)
// Re-register with _same_ proxy config
req, _ := http.NewRequest("PUT", "/v1/agent/service/register?token=root", jsonReader(sidecarProxy))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, 200, resp.Code, "body: %s", resp.Body.String())
},
wantWait: 200 * time.Millisecond,
wantCode: 200,
wantResp: expectedResponse,
},
{
// When we reload config, the agent pauses Anti-entropy, then clears all
// services (which causes their watch chans to be closed) before loading
// state from config/snapshot again). If we do that naively then we don't
// just get a spurios wakeup on the watch if the service didn't change,
// but we get it wakeup and then race with the reload and probably see no
2020-01-27 13:00:33 +00:00
// services and return a 404 error which is gross. This test exercises
// that - even though the registrations were from API not config, they are
// persisted and cleared/reloaded from snapshot which has same effect.
//
// The fix for this test is to allow the same mechanism that pauses
// Anti-entropy during reload to also pause the hash blocking loop so we
// don't resume until the state is reloaded and we get a chance to see if
// it actually changed or not.
name: "blocking fetch interrupted by reload shouldn't 404 - no change",
url: "/v1/agent/service/web-sidecar-proxy?wait=200ms&hash=" + expectedResponse.ContentHash,
updateFunc: func() {
time.Sleep(100 * time.Millisecond)
// Reload
require.NoError(t, a.reloadConfigInternal(a.Config))
},
// Should eventually timeout since there is no actual change
wantWait: 200 * time.Millisecond,
wantCode: 200,
wantResp: expectedResponse,
},
{
// As above but test actually altering the service with the config reload.
// This simulates the API registration being overridden by a different one
// on disk during reload.
name: "blocking fetch interrupted by reload shouldn't 404 - changes",
url: "/v1/agent/service/web-sidecar-proxy?wait=10m&hash=" + expectedResponse.ContentHash,
updateFunc: func() {
time.Sleep(100 * time.Millisecond)
// Reload
newConfig := *a.Config
newConfig.Services = append(newConfig.Services, &updatedProxy)
require.NoError(t, a.reloadConfigInternal(&newConfig))
},
wantWait: 100 * time.Millisecond,
wantCode: 200,
wantResp: &updatedResponse,
},
{
name: "err: non-existent proxy",
url: "/v1/agent/service/nope",
2022-01-05 20:18:08 +00:00
wantErr: fmt.Sprintf("unknown service ID: %s", structs.NewServiceID("nope", nil)),
},
{
name: "err: bad ACL for service",
url: "/v1/agent/service/web-sidecar-proxy",
// Limited token doesn't grant read to the service
tokenRules: `
key "" {
policy = "read"
}
`,
// Note that because we return ErrPermissionDenied and handle writing
// status at a higher level helper this actually gets a 200 in this test
// case so just assert that it was an error.
wantErr: "Permission denied",
},
{
name: "good ACL for service",
url: "/v1/agent/service/web-sidecar-proxy",
// Limited token doesn't grant read to the service
tokenRules: `
service "web-sidecar-proxy" {
policy = "read"
}
`,
wantCode: 200,
wantResp: expectedResponse,
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
// Register the basic service to ensure it's in a known state to start.
{
req, _ := http.NewRequest("PUT", "/v1/agent/service/register?token=root", jsonReader(sidecarProxy))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, 200, resp.Code, "body: %s", resp.Body.String())
}
req, _ := http.NewRequest("GET", tt.url, nil)
// Inject the root token for tests that don't care about ACL
2021-08-20 16:57:45 +00:00
token := "root"
if tt.tokenRules != "" {
// Create new token and use that.
token = testCreateToken(t, a, tt.tokenRules)
}
req.Header.Set("X-Consul-Token", token)
resp := httptest.NewRecorder()
if tt.updateFunc != nil {
go tt.updateFunc()
}
start := time.Now()
a.srv.h.ServeHTTP(resp, req)
elapsed := time.Since(start)
if tt.wantErr != "" {
require.Contains(t, strings.ToLower(resp.Body.String()), strings.ToLower(tt.wantErr))
}
if tt.wantCode != 0 {
require.Equal(t, tt.wantCode, resp.Code, "body: %s", resp.Body.String())
}
if tt.wantWait != 0 {
assert.True(t, elapsed >= tt.wantWait, "should have waited at least %s, "+
"took %s", tt.wantWait, elapsed)
} else {
assert.True(t, elapsed < 10*time.Millisecond, "should not have waited, "+
"took %s", elapsed)
}
if tt.wantResp != nil {
dec := json.NewDecoder(resp.Body)
val := &api.AgentService{}
err := dec.Decode(&val)
require.NoError(t, err)
assert.Equal(t, tt.wantResp, val)
assert.Equal(t, tt.wantResp.ContentHash, resp.Header().Get("X-Consul-ContentHash"))
}
})
}
}
func TestAgent_Checks(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
chk1 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "mysql",
Name: "mysql",
Interval: "30s",
Timeout: "5s",
Status: api.HealthPassing,
}
2017-08-28 12:17:13 +00:00
a.State.AddCheck(chk1, "")
req, _ := http.NewRequest("GET", "/v1/agent/checks", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
var val map[types.CheckID]*structs.HealthCheck
err := dec.Decode(&val)
if err != nil {
t.Fatalf("Err: %v", err)
}
if len(val) != 1 {
t.Fatalf("bad checks: %v", val)
}
if val["mysql"].Status != api.HealthPassing {
t.Fatalf("bad check: %v", val)
}
if val["mysql"].Node != chk1.Node {
t.Fatalf("bad check: %v", val)
}
if val["mysql"].Interval != chk1.Interval {
t.Fatalf("bad check: %v", val)
}
if val["mysql"].Timeout != chk1.Timeout {
t.Fatalf("bad check: %v", val)
}
}
func TestAgent_ChecksWithFilter(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
chk1 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "mysql",
Name: "mysql",
Status: api.HealthPassing,
}
a.State.AddCheck(chk1, "")
chk2 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "redis",
Name: "redis",
Status: api.HealthPassing,
}
a.State.AddCheck(chk2, "")
req, _ := http.NewRequest("GET", "/v1/agent/checks?filter="+url.QueryEscape("Name == `redis`"), nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
var val map[types.CheckID]*structs.HealthCheck
err := dec.Decode(&val)
require.NoError(t, err)
require.Len(t, val, 1)
_, ok := val["redis"]
require.True(t, ok)
}
func TestAgent_HealthServiceByID(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
service := &structs.NodeService{
ID: "mysql",
Service: "mysql",
}
serviceReq := AddServiceRequest{
Service: service,
chkTypes: nil,
persist: false,
token: "",
Source: ConfigSourceLocal,
}
if err := a.AddService(serviceReq); err != nil {
t.Fatalf("err: %v", err)
}
serviceReq.Service = &structs.NodeService{
ID: "mysql2",
Service: "mysql2",
}
if err := a.AddService(serviceReq); err != nil {
t.Fatalf("err: %v", err)
}
serviceReq.Service = &structs.NodeService{
ID: "mysql3",
Service: "mysql3",
}
if err := a.AddService(serviceReq); err != nil {
t.Fatalf("err: %v", err)
}
chk1 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "mysql",
Name: "mysql",
ServiceID: "mysql",
Status: api.HealthPassing,
}
err := a.State.AddCheck(chk1, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
chk2 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "mysql",
Name: "mysql",
ServiceID: "mysql",
Status: api.HealthPassing,
}
err = a.State.AddCheck(chk2, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
chk3 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "mysql2",
Name: "mysql2",
ServiceID: "mysql2",
Status: api.HealthPassing,
}
err = a.State.AddCheck(chk3, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
chk4 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "mysql2",
Name: "mysql2",
ServiceID: "mysql2",
Status: api.HealthWarning,
}
err = a.State.AddCheck(chk4, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
chk5 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "mysql3",
Name: "mysql3",
ServiceID: "mysql3",
Status: api.HealthMaint,
}
err = a.State.AddCheck(chk5, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
chk6 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "mysql3",
Name: "mysql3",
ServiceID: "mysql3",
Status: api.HealthCritical,
}
err = a.State.AddCheck(chk6, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
eval := func(t *testing.T, url string, expectedCode int, expected string) {
t.Helper()
t.Run("format=text", func(t *testing.T) {
t.Helper()
req, _ := http.NewRequest("GET", url+"?format=text", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
body := resp.Body.String()
if got, want := resp.Code, expectedCode; got != want {
t.Fatalf("returned bad status: expected %d, but had: %d", expectedCode, resp.Code)
}
if got, want := body, expected; got != want {
t.Fatalf("got body %q want %q", got, want)
}
})
t.Run("format=json", func(t *testing.T) {
req, _ := http.NewRequest("GET", url, nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if got, want := resp.Code, expectedCode; got != want {
t.Fatalf("returned bad status: expected %d, but had: %d", expectedCode, resp.Code)
}
dec := json.NewDecoder(resp.Body)
data := &api.AgentServiceChecksInfo{}
if err := dec.Decode(data); err != nil {
t.Fatalf("Cannot convert result from JSON: %v", err)
}
if resp.Code != http.StatusNotFound {
if data != nil && data.AggregatedStatus != expected {
t.Fatalf("got body %v want %v", data, expected)
}
}
})
}
t.Run("passing checks", func(t *testing.T) {
eval(t, "/v1/agent/health/service/id/mysql", http.StatusOK, "passing")
})
t.Run("warning checks", func(t *testing.T) {
eval(t, "/v1/agent/health/service/id/mysql2", http.StatusTooManyRequests, "warning")
})
t.Run("critical checks", func(t *testing.T) {
eval(t, "/v1/agent/health/service/id/mysql3", http.StatusServiceUnavailable, "critical")
})
t.Run("unknown serviceid", func(t *testing.T) {
eval(t, "/v1/agent/health/service/id/mysql1", http.StatusNotFound, fmt.Sprintf("ServiceId %s not found", structs.ServiceIDString("mysql1", nil)))
})
nodeCheck := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "diskCheck",
Name: "diskCheck",
Status: api.HealthCritical,
}
err = a.State.AddCheck(nodeCheck, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
t.Run("critical check on node", func(t *testing.T) {
eval(t, "/v1/agent/health/service/id/mysql", http.StatusServiceUnavailable, "critical")
})
err = a.State.RemoveCheck(nodeCheck.CompoundCheckID())
if err != nil {
t.Fatalf("Err: %v", err)
}
nodeCheck = &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "_node_maintenance",
Name: "_node_maintenance",
Status: api.HealthMaint,
}
err = a.State.AddCheck(nodeCheck, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
t.Run("maintenance check on node", func(t *testing.T) {
eval(t, "/v1/agent/health/service/id/mysql", http.StatusServiceUnavailable, "maintenance")
})
}
func TestAgent_HealthServiceByName(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
service := &structs.NodeService{
ID: "mysql1",
Service: "mysql-pool-r",
}
serviceReq := AddServiceRequest{
Service: service,
chkTypes: nil,
persist: false,
token: "",
Source: ConfigSourceLocal,
}
if err := a.AddService(serviceReq); err != nil {
t.Fatalf("err: %v", err)
}
serviceReq.Service = &structs.NodeService{
ID: "mysql2",
Service: "mysql-pool-r",
}
if err := a.AddService(serviceReq); err != nil {
t.Fatalf("err: %v", err)
}
serviceReq.Service = &structs.NodeService{
ID: "mysql3",
Service: "mysql-pool-rw",
}
if err := a.AddService(serviceReq); err != nil {
t.Fatalf("err: %v", err)
}
serviceReq.Service = &structs.NodeService{
ID: "mysql4",
Service: "mysql-pool-rw",
}
if err := a.AddService(serviceReq); err != nil {
t.Fatalf("err: %v", err)
}
serviceReq.Service = &structs.NodeService{
ID: "httpd1",
Service: "httpd",
}
if err := a.AddService(serviceReq); err != nil {
t.Fatalf("err: %v", err)
}
serviceReq.Service = &structs.NodeService{
ID: "httpd2",
Service: "httpd",
}
if err := a.AddService(serviceReq); err != nil {
t.Fatalf("err: %v", err)
}
chk1 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "mysql1",
Name: "mysql1",
ServiceID: "mysql1",
ServiceName: "mysql-pool-r",
Status: api.HealthPassing,
}
err := a.State.AddCheck(chk1, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
chk2 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "mysql1",
Name: "mysql1",
ServiceID: "mysql1",
ServiceName: "mysql-pool-r",
Status: api.HealthWarning,
}
err = a.State.AddCheck(chk2, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
chk3 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "mysql2",
Name: "mysql2",
ServiceID: "mysql2",
ServiceName: "mysql-pool-r",
Status: api.HealthPassing,
}
err = a.State.AddCheck(chk3, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
chk4 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "mysql2",
Name: "mysql2",
ServiceID: "mysql2",
ServiceName: "mysql-pool-r",
Status: api.HealthCritical,
}
err = a.State.AddCheck(chk4, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
chk5 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "mysql3",
Name: "mysql3",
ServiceID: "mysql3",
ServiceName: "mysql-pool-rw",
Status: api.HealthWarning,
}
err = a.State.AddCheck(chk5, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
chk6 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "mysql4",
Name: "mysql4",
ServiceID: "mysql4",
ServiceName: "mysql-pool-rw",
Status: api.HealthPassing,
}
err = a.State.AddCheck(chk6, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
chk7 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "httpd1",
Name: "httpd1",
ServiceID: "httpd1",
ServiceName: "httpd",
Status: api.HealthPassing,
}
err = a.State.AddCheck(chk7, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
chk8 := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "httpd2",
Name: "httpd2",
ServiceID: "httpd2",
ServiceName: "httpd",
Status: api.HealthPassing,
}
err = a.State.AddCheck(chk8, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
eval := func(t *testing.T, url string, expectedCode int, expected string) {
t.Helper()
t.Run("format=text", func(t *testing.T) {
t.Helper()
req, _ := http.NewRequest("GET", url+"?format=text", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if got, want := resp.Code, expectedCode; got != want {
t.Fatalf("returned bad status: %d. Body: %q", resp.Code, resp.Body.String())
}
if got, want := resp.Body.String(), expected; got != want {
t.Fatalf("got body %q want %q", got, want)
}
})
t.Run("format=json", func(t *testing.T) {
t.Helper()
req, _ := http.NewRequest("GET", url, nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
data := make([]*api.AgentServiceChecksInfo, 0)
if err := dec.Decode(&data); err != nil {
t.Fatalf("Cannot convert result from JSON: %v", err)
}
if got, want := resp.Code, expectedCode; got != want {
t.Fatalf("returned bad code: %d. Body: %#v", resp.Code, data)
}
if resp.Code != http.StatusNotFound {
matched := false
for _, d := range data {
if d.AggregatedStatus == expected {
matched = true
break
}
}
if !matched {
t.Fatalf("got wrong status, wanted %#v", expected)
}
}
})
}
t.Run("passing checks", func(t *testing.T) {
eval(t, "/v1/agent/health/service/name/httpd", http.StatusOK, "passing")
})
t.Run("warning checks", func(t *testing.T) {
eval(t, "/v1/agent/health/service/name/mysql-pool-rw", http.StatusTooManyRequests, "warning")
})
t.Run("critical checks", func(t *testing.T) {
eval(t, "/v1/agent/health/service/name/mysql-pool-r", http.StatusServiceUnavailable, "critical")
})
t.Run("unknown serviceName", func(t *testing.T) {
eval(t, "/v1/agent/health/service/name/test", http.StatusNotFound, "ServiceName test Not Found")
})
nodeCheck := &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "diskCheck",
Name: "diskCheck",
Status: api.HealthCritical,
}
err = a.State.AddCheck(nodeCheck, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
t.Run("critical check on node", func(t *testing.T) {
eval(t, "/v1/agent/health/service/name/mysql-pool-r", http.StatusServiceUnavailable, "critical")
})
err = a.State.RemoveCheck(nodeCheck.CompoundCheckID())
if err != nil {
t.Fatalf("Err: %v", err)
}
nodeCheck = &structs.HealthCheck{
Node: a.Config.NodeName,
CheckID: "_node_maintenance",
Name: "_node_maintenance",
Status: api.HealthMaint,
}
err = a.State.AddCheck(nodeCheck, "")
if err != nil {
t.Fatalf("Err: %v", err)
}
t.Run("maintenance check on node", func(t *testing.T) {
eval(t, "/v1/agent/health/service/name/mysql-pool-r", http.StatusServiceUnavailable, "maintenance")
})
}
func TestAgent_HealthServicesACLEnforcement(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfigWithParams(nil))
defer a.Shutdown()
service := &structs.NodeService{
ID: "mysql1",
Service: "mysql",
}
serviceReq := AddServiceRequest{
Service: service,
chkTypes: nil,
persist: false,
token: "",
Source: ConfigSourceLocal,
}
require.NoError(t, a.AddService(serviceReq))
serviceReq.Service = &structs.NodeService{
ID: "foo1",
Service: "foo",
}
require.NoError(t, a.AddService(serviceReq))
// no token
t.Run("no-token-health-by-id", func(t *testing.T) {
req, err := http.NewRequest("GET", "/v1/agent/health/service/id/mysql1", nil)
require.NoError(t, err)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("no-token-health-by-name", func(t *testing.T) {
req, err := http.NewRequest("GET", "/v1/agent/health/service/name/mysql", nil)
require.NoError(t, err)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("root-token-health-by-id", func(t *testing.T) {
req, err := http.NewRequest("GET", "/v1/agent/health/service/id/foo1", nil)
require.NoError(t, err)
req.Header.Add("X-Consul-Token", TestDefaultInitialManagementToken)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
t.Run("root-token-health-by-name", func(t *testing.T) {
req, err := http.NewRequest("GET", "/v1/agent/health/service/name/foo", nil)
require.NoError(t, err)
req.Header.Add("X-Consul-Token", TestDefaultInitialManagementToken)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
}
func TestAgent_Checks_ACLFilter(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
checks := structs.HealthChecks{
{
Node: a.Config.NodeName,
CheckID: "web",
ServiceName: "web",
Status: api.HealthPassing,
},
{
Node: a.Config.NodeName,
CheckID: "api",
ServiceName: "api",
Status: api.HealthPassing,
},
}
for _, c := range checks {
a.State.AddCheck(c, "")
}
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("GET", "/v1/agent/checks", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
val := make(map[types.CheckID]*structs.HealthCheck)
if err := dec.Decode(&val); err != nil {
t.Fatalf("Err: %v", err)
}
require.Len(t, val, 0)
require.Empty(t, resp.Header().Get("X-Consul-Results-Filtered-By-ACLs"))
})
t.Run("limited token", func(t *testing.T) {
token := testCreateToken(t, a, fmt.Sprintf(`
service "web" {
policy = "read"
}
node "%s" {
policy = "read"
}
`, a.Config.NodeName))
req := httptest.NewRequest("GET", fmt.Sprintf("/v1/agent/checks?token=%s", token), nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
var val map[types.CheckID]*structs.HealthCheck
if err := dec.Decode(&val); err != nil {
t.Fatalf("Err: %v", err)
}
require.Len(t, val, 1)
require.NotEmpty(t, resp.Header().Get("X-Consul-Results-Filtered-By-ACLs"))
})
t.Run("root token", func(t *testing.T) {
req, _ := http.NewRequest("GET", "/v1/agent/checks?token=root", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
val := make(map[types.CheckID]*structs.HealthCheck)
if err := dec.Decode(&val); err != nil {
t.Fatalf("Err: %v", err)
}
require.Len(t, val, 2)
require.Empty(t, resp.Header().Get("X-Consul-Results-Filtered-By-ACLs"))
})
}
func TestAgent_Self(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
cases := map[string]struct {
hcl string
expectXDS bool
}{
"normal": {
hcl: `
node_meta {
somekey = "somevalue"
}
`,
expectXDS: true,
},
"no grpc": {
hcl: `
node_meta {
somekey = "somevalue"
}
ports = {
grpc = -1
}
`,
expectXDS: false,
},
}
for name, tc := range cases {
tc := tc
t.Run(name, func(t *testing.T) {
a := NewTestAgent(t, tc.hcl)
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
req, _ := http.NewRequest("GET", "/v1/agent/self", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
val := &Self{}
require.NoError(t, dec.Decode(val))
require.Equal(t, a.Config.SerfPortLAN, int(val.Member.Port))
require.Equal(t, a.Config.SerfPortLAN, int(val.DebugConfig["SerfPortLAN"].(float64)))
cs, err := a.GetLANCoordinate()
require.NoError(t, err)
require.Equal(t, cs[a.config.SegmentName], val.Coord)
delete(val.Meta, structs.MetaSegmentKey) // Added later, not in config.
require.Equal(t, a.config.NodeMeta, val.Meta)
if tc.expectXDS {
require.NotNil(t, val.XDS, "xds component missing when gRPC is enabled")
require.Equal(t,
map[string][]string{"envoy": proxysupport.EnvoyVersions},
val.XDS.SupportedProxies,
)
} else {
require.Nil(t, val.XDS, "xds component should be missing when gRPC is disabled")
}
})
}
}
func TestAgent_Self_ACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("GET", "/v1/agent/self", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("agent recovery token", func(t *testing.T) {
req, _ := http.NewRequest("GET", "/v1/agent/self?token=towel", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
t.Run("read-only token", func(t *testing.T) {
ro := createACLTokenWithAgentReadPolicy(t, a.srv)
req, _ := http.NewRequest("GET", fmt.Sprintf("/v1/agent/self?token=%s", ro), nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
}
2017-08-08 20:05:38 +00:00
func TestAgent_Metrics_ACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
2017-08-08 20:05:38 +00:00
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
2017-08-08 20:05:38 +00:00
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
2017-08-08 20:05:38 +00:00
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("GET", "/v1/agent/metrics", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
2017-08-08 20:05:38 +00:00
})
t.Run("agent recovery token", func(t *testing.T) {
2017-08-08 20:05:38 +00:00
req, _ := http.NewRequest("GET", "/v1/agent/metrics?token=towel", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
2017-08-08 20:05:38 +00:00
})
t.Run("read-only token", func(t *testing.T) {
ro := createACLTokenWithAgentReadPolicy(t, a.srv)
2017-08-08 20:05:38 +00:00
req, _ := http.NewRequest("GET", fmt.Sprintf("/v1/agent/metrics?token=%s", ro), nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
2017-08-08 20:05:38 +00:00
})
}
2021-07-15 23:04:24 +00:00
func TestHTTPHandlers_AgentMetricsStream_ACLDeny(t *testing.T) {
bd := BaseDeps{}
bd.Tokens = new(tokenStore.Store)
sink := metrics.NewInmemSink(30*time.Millisecond, time.Second)
bd.MetricsHandler = sink
d := fakeResolveTokenDelegate{authorizer: acl.DenyAll()}
agent := &Agent{
baseDeps: bd,
delegate: d,
tokens: bd.Tokens,
config: &config.RuntimeConfig{NodeName: "the-node"},
logger: hclog.NewInterceptLogger(nil),
}
h := HTTPHandlers{agent: agent, denylist: NewDenylist(nil)}
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
resp := httptest.NewRecorder()
req, err := http.NewRequestWithContext(ctx, http.MethodGet, "/v1/agent/metrics/stream", nil)
require.NoError(t, err)
handle := h.handler(false)
handle.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
require.Contains(t, resp.Body.String(), "Permission denied")
}
func TestHTTPHandlers_AgentMetricsStream(t *testing.T) {
bd := BaseDeps{}
bd.Tokens = new(tokenStore.Store)
sink := metrics.NewInmemSink(20*time.Millisecond, time.Second)
bd.MetricsHandler = sink
2021-07-30 17:16:38 +00:00
d := fakeResolveTokenDelegate{authorizer: acl.ManageAll()}
2021-07-15 23:04:24 +00:00
agent := &Agent{
baseDeps: bd,
delegate: d,
tokens: bd.Tokens,
config: &config.RuntimeConfig{NodeName: "the-node"},
logger: hclog.NewInterceptLogger(nil),
}
h := HTTPHandlers{agent: agent, denylist: NewDenylist(nil)}
ctx, cancel := context.WithTimeout(context.Background(), 60*time.Millisecond)
defer cancel()
// produce some metrics
go func() {
for ctx.Err() == nil {
sink.SetGauge([]string{"the-key"}, 12)
time.Sleep(5 * time.Millisecond)
}
}()
resp := httptest.NewRecorder()
req, err := http.NewRequestWithContext(ctx, http.MethodGet, "/v1/agent/metrics/stream", nil)
require.NoError(t, err)
handle := h.handler(false)
handle.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
decoder := json.NewDecoder(resp.Body)
var summary metrics.MetricsSummary
err = decoder.Decode(&summary)
require.NoError(t, err)
expected := []metrics.GaugeValue{
{Name: "the-key", Value: 12, DisplayLabels: map[string]string{}},
}
require.Equal(t, expected, summary.Gauges)
// There should be at least two intervals worth of metrics
err = decoder.Decode(&summary)
require.NoError(t, err)
require.Equal(t, expected, summary.Gauges)
}
type fakeResolveTokenDelegate struct {
delegate
authorizer acl.Authorizer
}
func (f fakeResolveTokenDelegate) ResolveTokenAndDefaultMeta(_ string, _ *structs.EnterpriseMeta, _ *acl.AuthorizerContext) (consul.ACLResolveResult, error) {
return consul.ACLResolveResult{Authorizer: f.authorizer}, nil
2021-07-15 23:04:24 +00:00
}
func TestAgent_Reload(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
dc1 := "dc1"
a := NewTestAgent(t, `
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
services = [
{
name = "redis"
}
]
watches = [
{
datacenter = "`+dc1+`"
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
type = "key"
key = "test"
handler = "true"
}
]
limits = {
rpc_rate=1
rpc_max_burst=100
}
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
`)
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, dc1)
if a.State.Service(structs.NewServiceID("redis", nil)) == nil {
2017-08-28 12:17:13 +00:00
t.Fatal("missing redis service")
}
cfg2 := TestConfig(testutil.Logger(t), config.FileSource{
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
Name: "reload",
Format: "hcl",
Data: `
data_dir = "` + a.Config.DataDir + `"
node_id = "` + string(a.Config.NodeID) + `"
node_name = "` + a.Config.NodeName + `"
services = [
{
name = "redis-reloaded"
}
]
limits = {
rpc_rate=2
rpc_max_burst=200
}
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
`,
})
shim := &delegateConfigReloadShim{delegate: a.delegate}
a.delegate = shim
if err := a.reloadConfigInternal(cfg2); err != nil {
t.Fatalf("got error %v want nil", err)
}
if a.State.Service(structs.NewServiceID("redis-reloaded", nil)) == nil {
2017-08-28 12:17:13 +00:00
t.Fatal("missing redis-reloaded service")
}
require.Equal(t, rate.Limit(2), shim.newCfg.RPCRateLimit)
require.Equal(t, 200, shim.newCfg.RPCMaxBurst)
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
for _, wp := range a.watchPlans {
if !wp.IsStopped() {
t.Fatalf("Reloading configs should stop watch plans of the previous configuration")
}
}
}
type delegateConfigReloadShim struct {
delegate
newCfg consul.ReloadableConfig
}
func (s *delegateConfigReloadShim) ReloadConfig(cfg consul.ReloadableConfig) error {
s.newCfg = cfg
return s.delegate.ReloadConfig(cfg)
}
// TestAgent_ReloadDoesNotTriggerWatch Ensure watches not triggered after reload
// see https://github.com/hashicorp/consul/issues/7446
func TestAgent_ReloadDoesNotTriggerWatch(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
dc1 := "dc1"
tmpFileRaw, err := ioutil.TempFile("", "rexec")
require.NoError(t, err)
tmpFile := tmpFileRaw.Name()
defer os.Remove(tmpFile)
handlerShell := fmt.Sprintf("(cat ; echo CONSUL_INDEX $CONSUL_INDEX) | tee '%s.atomic' ; mv '%s.atomic' '%s'", tmpFile, tmpFile, tmpFile)
a := NewTestAgent(t, `
services = [
{
name = "redis"
checks = [
{
id = "red-is-dead"
ttl = "30s"
notes = "initial check"
}
]
}
]
watches = [
{
datacenter = "`+dc1+`"
type = "service"
service = "redis"
args = ["bash", "-c", "`+handlerShell+`"]
}
]
`)
checkID := structs.NewCheckID("red-is-dead", nil)
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, dc1)
require.NoError(t, a.updateTTLCheck(checkID, api.HealthPassing, "testing-agent-reload-001"))
checkStr := func(r *retry.R, evaluator func(string) error) {
t.Helper()
contentsStr := ""
// Wait for watch to be populated
for i := 1; i < 7; i++ {
contents, err := ioutil.ReadFile(tmpFile)
if err != nil {
t.Fatalf("should be able to read file, but had: %#v", err)
}
contentsStr = string(contents)
if contentsStr != "" {
break
}
time.Sleep(time.Duration(i) * time.Second)
testutil.Logger(t).Info("Watch not yet populated, retrying")
}
if err := evaluator(contentsStr); err != nil {
r.Errorf("ERROR: Test failing: %s", err)
}
}
ensureNothingCritical := func(r *retry.R, mustContain string) {
t.Helper()
eval := func(contentsStr string) error {
if strings.Contains(contentsStr, "critical") {
return fmt.Errorf("MUST NOT contain critical:= %s", contentsStr)
}
if !strings.Contains(contentsStr, mustContain) {
return fmt.Errorf("MUST contain '%s' := %s", mustContain, contentsStr)
}
return nil
}
checkStr(r, eval)
}
retriesWithDelay := func() *retry.Counter {
return &retry.Counter{Count: 10, Wait: 1 * time.Second}
}
retry.RunWith(retriesWithDelay(), t, func(r *retry.R) {
testutil.Logger(t).Info("Consul is now ready")
// it should contain the output
checkStr(r, func(contentStr string) error {
if contentStr == "[]" {
return fmt.Errorf("Consul is still starting up")
}
return nil
})
})
retry.RunWith(retriesWithDelay(), t, func(r *retry.R) {
ensureNothingCritical(r, "testing-agent-reload-001")
})
// Let's take almost the same config
cfg2 := TestConfig(testutil.Logger(t), config.FileSource{
Name: "reload",
Format: "hcl",
Data: `
data_dir = "` + a.Config.DataDir + `"
node_id = "` + string(a.Config.NodeID) + `"
node_name = "` + a.Config.NodeName + `"
services = [
{
name = "redis"
checks = [
{
id = "red-is-dead"
ttl = "30s"
notes = "initial check"
}
]
}
]
watches = [
{
datacenter = "` + dc1 + `"
type = "service"
service = "redis"
args = ["bash", "-c", "` + handlerShell + `"]
}
]
`,
})
justOnce := func() *retry.Counter {
return &retry.Counter{Count: 1, Wait: 25 * time.Millisecond}
}
retry.RunWith(justOnce(), t, func(r *retry.R) {
// We check that reload does not go to critical
ensureNothingCritical(r, "red-is-dead")
if err := a.reloadConfigInternal(cfg2); err != nil {
t.Fatalf("got error %v want nil", err)
}
// We check that reload does not go to critical
ensureNothingCritical(r, "red-is-dead")
ensureNothingCritical(r, "testing-agent-reload-001")
require.NoError(t, a.updateTTLCheck(checkID, api.HealthPassing, "testing-agent-reload-002"))
ensureNothingCritical(r, "red-is-dead")
})
}
func TestAgent_Reload_ACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/reload", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("read-only token", func(t *testing.T) {
ro := createACLTokenWithAgentReadPolicy(t, a.srv)
req, _ := http.NewRequest("PUT", fmt.Sprintf("/v1/agent/reload?token=%s", ro), nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
// This proves we call the ACL function, and we've got the other reload
// test to prove we do the reload, which should be sufficient.
// The reload logic is a little complex to set up so isn't worth
// repeating again here.
}
func TestAgent_Members(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
req, _ := http.NewRequest("GET", "/v1/agent/members", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
val := make([]serf.Member, 0)
if err := dec.Decode(&val); err != nil {
t.Fatalf("Err: %v", err)
}
if len(val) == 0 {
t.Fatalf("bad members: %v", val)
}
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
if int(val[0].Port) != a.Config.SerfPortLAN {
t.Fatalf("not lan: %v", val)
}
}
func TestAgent_Members_WAN(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
req, _ := http.NewRequest("GET", "/v1/agent/members?wan=true", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
val := make([]serf.Member, 0)
if err := dec.Decode(&val); err != nil {
t.Fatalf("Err: %v", err)
}
if len(val) == 0 {
t.Fatalf("bad members: %v", val)
}
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
if int(val[0].Port) != a.Config.SerfPortWAN {
t.Fatalf("not wan: %v", val)
}
}
func TestAgent_Members_ACLFilter(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
// Start 2 agents and join them together.
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
b := NewTestAgent(t, TestACLConfig())
defer b.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
testrpc.WaitForLeader(t, b.RPC, "dc1")
joinPath := fmt.Sprintf("/v1/agent/join/127.0.0.1:%d?token=root", b.Config.SerfPortLAN)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, httptest.NewRequest(http.MethodPut, joinPath, nil))
require.Equal(t, http.StatusOK, resp.Code)
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("GET", "/v1/agent/members", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
val := make([]serf.Member, 0)
if err := dec.Decode(&val); err != nil {
t.Fatalf("Err: %v", err)
}
require.Len(t, val, 0)
require.Empty(t, resp.Header().Get("X-Consul-Results-Filtered-By-ACLs"))
})
t.Run("limited token", func(t *testing.T) {
token := testCreateToken(t, a, fmt.Sprintf(`
node "%s" {
policy = "read"
}
`, b.Config.NodeName))
req := httptest.NewRequest("GET", fmt.Sprintf("/v1/agent/members?token=%s", token), nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
val := make([]serf.Member, 0)
if err := dec.Decode(&val); err != nil {
t.Fatalf("Err: %v", err)
}
require.Len(t, val, 1)
require.NotEmpty(t, resp.Header().Get("X-Consul-Results-Filtered-By-ACLs"))
})
t.Run("root token", func(t *testing.T) {
req, _ := http.NewRequest("GET", "/v1/agent/members?token=root", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
val := make([]serf.Member, 0)
if err := dec.Decode(&val); err != nil {
t.Fatalf("Err: %v", err)
}
require.Len(t, val, 2)
require.Empty(t, resp.Header().Get("X-Consul-Results-Filtered-By-ACLs"))
})
}
func TestAgent_Join(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a1 := NewTestAgent(t, "")
defer a1.Shutdown()
a2 := NewTestAgent(t, "")
defer a2.Shutdown()
testrpc.WaitForLeader(t, a1.RPC, "dc1")
testrpc.WaitForLeader(t, a2.RPC, "dc1")
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
addr := fmt.Sprintf("127.0.0.1:%d", a2.Config.SerfPortLAN)
req, _ := http.NewRequest("PUT", fmt.Sprintf("/v1/agent/join/%s", addr), nil)
resp := httptest.NewRecorder()
a1.srv.h.ServeHTTP(resp, req)
if len(a1.LANMembersInAgentPartition()) != 2 {
t.Fatalf("should have 2 members")
}
retry.Run(t, func(r *retry.R) {
if got, want := len(a2.LANMembersInAgentPartition()), 2; got != want {
r.Fatalf("got %d LAN members want %d", got, want)
}
})
}
func TestAgent_Join_WAN(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a1 := NewTestAgent(t, "")
defer a1.Shutdown()
a2 := NewTestAgent(t, "")
defer a2.Shutdown()
testrpc.WaitForLeader(t, a1.RPC, "dc1")
testrpc.WaitForLeader(t, a2.RPC, "dc1")
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
addr := fmt.Sprintf("127.0.0.1:%d", a2.Config.SerfPortWAN)
req, _ := http.NewRequest("PUT", fmt.Sprintf("/v1/agent/join/%s?wan=true", addr), nil)
resp := httptest.NewRecorder()
a1.srv.h.ServeHTTP(resp, req)
if len(a1.WANMembers()) != 2 {
t.Fatalf("should have 2 members")
}
retry.Run(t, func(r *retry.R) {
if got, want := len(a2.WANMembers()), 2; got != want {
r.Fatalf("got %d WAN members want %d", got, want)
}
})
}
func TestAgent_Join_ACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a1 := NewTestAgent(t, TestACLConfig())
defer a1.Shutdown()
a2 := NewTestAgent(t, "")
defer a2.Shutdown()
testrpc.WaitForLeader(t, a1.RPC, "dc1")
testrpc.WaitForLeader(t, a2.RPC, "dc1")
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
addr := fmt.Sprintf("127.0.0.1:%d", a2.Config.SerfPortLAN)
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", fmt.Sprintf("/v1/agent/join/%s", addr), nil)
resp := httptest.NewRecorder()
a1.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("agent recovery token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", fmt.Sprintf("/v1/agent/join/%s?token=towel", addr), nil)
resp := httptest.NewRecorder()
a1.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
t.Run("read-only token", func(t *testing.T) {
ro := createACLTokenWithAgentReadPolicy(t, a1.srv)
req, _ := http.NewRequest("PUT", fmt.Sprintf("/v1/agent/join/%s?token=%s", addr, ro), nil)
resp := httptest.NewRecorder()
a1.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
}
type mockNotifier struct{ s string }
func (n *mockNotifier) Notify(state string) error {
n.s = state
return nil
}
func TestAgent_JoinLANNotify(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a1 := NewTestAgent(t, "")
defer a1.Shutdown()
testrpc.WaitForLeader(t, a1.RPC, "dc1")
a2 := NewTestAgent(t, `
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
server = false
bootstrap = false
`)
defer a2.Shutdown()
notif := &mockNotifier{}
a1.joinLANNotifier = notif
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
addr := fmt.Sprintf("127.0.0.1:%d", a2.Config.SerfPortLAN)
_, err := a1.JoinLAN([]string{addr}, nil)
if err != nil {
t.Fatalf("err: %v", err)
}
if got, want := notif.s, "READY=1"; got != want {
t.Fatalf("got joinLAN notification %q want %q", got, want)
}
}
func TestAgent_Leave(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a1 := NewTestAgent(t, "")
defer a1.Shutdown()
testrpc.WaitForLeader(t, a1.RPC, "dc1")
a2 := NewTestAgent(t, `
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
server = false
bootstrap = false
`)
defer a2.Shutdown()
// Join first
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
addr := fmt.Sprintf("127.0.0.1:%d", a2.Config.SerfPortLAN)
_, err := a1.JoinLAN([]string{addr}, nil)
if err != nil {
t.Fatalf("err: %v", err)
}
// Graceful leave now
req, _ := http.NewRequest("PUT", "/v1/agent/leave", nil)
resp := httptest.NewRecorder()
a2.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
retry.Run(t, func(r *retry.R) {
m := a1.LANMembersInAgentPartition()
if got, want := m[1].Status, serf.StatusLeft; got != want {
r.Fatalf("got status %q want %q", got, want)
}
})
}
func TestAgent_Leave_ACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/leave", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("read-only token", func(t *testing.T) {
ro := createACLTokenWithAgentReadPolicy(t, a.srv)
req, _ := http.NewRequest("PUT", fmt.Sprintf("/v1/agent/leave?token=%s", ro), nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
// this sub-test will change the state so that there is no leader.
// it must therefore be the last one in this list.
t.Run("agent recovery token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/leave?token=towel", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
}
func TestAgent_ForceLeave(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a1 := NewTestAgent(t, "")
defer a1.Shutdown()
a2 := NewTestAgent(t, "")
testrpc.WaitForLeader(t, a1.RPC, "dc1")
testrpc.WaitForLeader(t, a2.RPC, "dc1")
// Join first
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
addr := fmt.Sprintf("127.0.0.1:%d", a2.Config.SerfPortLAN)
_, err := a1.JoinLAN([]string{addr}, nil)
if err != nil {
t.Fatalf("err: %v", err)
}
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
// this test probably needs work
a2.Shutdown()
// Wait for agent being marked as failed, so we wait for full shutdown of Agent
retry.Run(t, func(r *retry.R) {
m := a1.LANMembersInAgentPartition()
if got, want := m[1].Status, serf.StatusFailed; got != want {
r.Fatalf("got status %q want %q", got, want)
}
})
// Force leave now
req, _ := http.NewRequest("PUT", fmt.Sprintf("/v1/agent/force-leave/%s", a2.Config.NodeName), nil)
resp := httptest.NewRecorder()
a1.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
retry.Run(t, func(r *retry.R) {
m := a1.LANMembersInAgentPartition()
if got, want := m[1].Status, serf.StatusLeft; got != want {
r.Fatalf("got status %q want %q", got, want)
}
})
}
func TestOpenMetricsMimeTypeHeaders(t *testing.T) {
t.Parallel()
assert.False(t, acceptsOpenMetricsMimeType(""))
assert.False(t, acceptsOpenMetricsMimeType(";;;"))
assert.False(t, acceptsOpenMetricsMimeType(",,,"))
assert.False(t, acceptsOpenMetricsMimeType("text/plain"))
assert.True(t, acceptsOpenMetricsMimeType("text/plain;version=0.4.0,"))
assert.True(t, acceptsOpenMetricsMimeType("text/plain;version=0.4.0;q=1,*/*;q=0.1"))
assert.True(t, acceptsOpenMetricsMimeType("text/plain ; version=0.4.0"))
assert.True(t, acceptsOpenMetricsMimeType("*/*, application/openmetrics-text ;"))
assert.True(t, acceptsOpenMetricsMimeType("*/*, application/openmetrics-text ;q=1"))
assert.True(t, acceptsOpenMetricsMimeType("application/openmetrics-text, text/plain;version=0.4.0"))
assert.True(t, acceptsOpenMetricsMimeType("application/openmetrics-text; version=0.0.1,text/plain;version=0.0.4;q=0.5,*/*;q=0.1"))
}
func TestAgent_ForceLeave_ACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
uri := fmt.Sprintf("/v1/agent/force-leave/%s", a.Config.NodeName)
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", uri, nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("agent recovery token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", uri+"?token=towel", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("read-only token", func(t *testing.T) {
ro := createACLTokenWithAgentReadPolicy(t, a.srv)
req, _ := http.NewRequest("PUT", fmt.Sprintf(uri+"?token=%s", ro), nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("operator write token", func(t *testing.T) {
// Create an ACL with operator read permissions.
2021-08-20 16:57:45 +00:00
rules := `
operator = "write"
`
opToken := testCreateToken(t, a, rules)
req, _ := http.NewRequest("PUT", fmt.Sprintf(uri+"?token=%s", opToken), nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
}
func TestAgent_ForceLeavePrune(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a1 := StartTestAgent(t, TestAgent{Name: "Agent1"})
defer a1.Shutdown()
a2 := StartTestAgent(t, TestAgent{Name: "Agent2"})
testrpc.WaitForLeader(t, a1.RPC, "dc1")
testrpc.WaitForLeader(t, a2.RPC, "dc1")
// Join first
addr := fmt.Sprintf("127.0.0.1:%d", a2.Config.SerfPortLAN)
_, err := a1.JoinLAN([]string{addr}, nil)
if err != nil {
t.Fatalf("err: %v", err)
}
// this test probably needs work
a2.Shutdown()
// Wait for agent being marked as failed, so we wait for full shutdown of Agent
retry.Run(t, func(r *retry.R) {
m := a1.LANMembersInAgentPartition()
for _, member := range m {
if member.Name == a2.Config.NodeName {
if member.Status != serf.StatusFailed {
r.Fatalf("got status %q want %q", member.Status, serf.StatusFailed)
}
}
}
})
// Force leave now
req, _ := http.NewRequest("PUT", fmt.Sprintf("/v1/agent/force-leave/%s?prune=true", a2.Config.NodeName), nil)
resp := httptest.NewRecorder()
a1.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
retry.Run(t, func(r *retry.R) {
m := len(a1.LANMembersInAgentPartition())
if m != 1 {
r.Fatalf("want one member, got %v", m)
}
})
}
func TestAgent_ForceLeavePrune_WAN(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a1 := StartTestAgent(t, TestAgent{Name: "dc1", HCL: `
datacenter = "dc1"
primary_datacenter = "dc1"
gossip_wan {
probe_interval = "50ms"
suspicion_mult = 2
}
`})
defer a1.Shutdown()
a2 := StartTestAgent(t, TestAgent{Name: "dc2", HCL: `
datacenter = "dc2"
primary_datacenter = "dc1"
`})
defer a2.Shutdown()
testrpc.WaitForLeader(t, a1.RPC, "dc1")
testrpc.WaitForLeader(t, a2.RPC, "dc2")
// Wait for the WAN join.
addr := fmt.Sprintf("127.0.0.1:%d", a1.Config.SerfPortWAN)
_, err := a2.JoinWAN([]string{addr})
require.NoError(t, err)
testrpc.WaitForLeader(t, a1.RPC, "dc2")
testrpc.WaitForLeader(t, a2.RPC, "dc1")
retry.Run(t, func(r *retry.R) {
require.Len(r, a1.WANMembers(), 2)
require.Len(r, a2.WANMembers(), 2)
})
wanNodeName_a2 := a2.Config.NodeName + ".dc2"
// Shutdown and wait for agent being marked as failed, so we wait for full
// shutdown of Agent.
require.NoError(t, a2.Shutdown())
retry.Run(t, func(r *retry.R) {
m := a1.WANMembers()
for _, member := range m {
if member.Name == wanNodeName_a2 {
if member.Status != serf.StatusFailed {
r.Fatalf("got status %q want %q", member.Status, serf.StatusFailed)
}
}
}
})
// Force leave now
req, err := http.NewRequest("PUT", fmt.Sprintf("/v1/agent/force-leave/%s?prune=1&wan=1", wanNodeName_a2), nil)
require.NoError(t, err)
resp := httptest.NewRecorder()
a1.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code, resp.Body.String())
retry.Run(t, func(r *retry.R) {
require.Len(r, a1.WANMembers(), 1)
})
}
func TestAgent_RegisterCheck(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
args := &structs.CheckDefinition{
Name: "test",
TTL: 15 * time.Second,
}
req, _ := http.NewRequest("PUT", "/v1/agent/check/register?token=abc123", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
// Ensure we have a check mapping
checkID := structs.NewCheckID("test", nil)
if existing := a.State.Check(checkID); existing == nil {
t.Fatalf("missing test check")
}
if _, ok := a.checkTTLs[checkID]; !ok {
t.Fatalf("missing test check ttl")
}
2015-04-28 20:06:02 +00:00
// Ensure the token was configured
2017-08-28 12:17:13 +00:00
if token := a.State.CheckToken(checkID); token == "" {
2015-04-28 20:06:02 +00:00
t.Fatalf("missing token")
}
2015-05-11 23:48:10 +00:00
// By default, checks start in critical state.
state := a.State.Check(checkID)
if state.Status != api.HealthCritical {
t.Fatalf("bad: %v", state)
}
}
// This verifies all the forms of the new args-style check that we need to
// support as a result of https://github.com/hashicorp/consul/issues/3587.
func TestAgent_RegisterCheck_Scripts(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, `
enable_script_checks = true
`)
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
tests := []struct {
name string
check map[string]interface{}
}{
{
"== Consul 1.0.0",
map[string]interface{}{
"Name": "test",
"Interval": "2s",
"ScriptArgs": []string{"true"},
},
},
{
"> Consul 1.0.0 (fixup)",
map[string]interface{}{
"Name": "test",
"Interval": "2s",
"script_args": []string{"true"},
},
},
{
"> Consul 1.0.0",
map[string]interface{}{
"Name": "test",
"Interval": "2s",
"Args": []string{"true"},
},
},
}
for _, tt := range tests {
t.Run(tt.name+" as node check", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/register", jsonReader(tt.check))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if resp.Code != http.StatusOK {
t.Fatalf("bad: %d", resp.Code)
}
})
t.Run(tt.name+" as top-level service check", func(t *testing.T) {
args := map[string]interface{}{
"Name": "a",
"Port": 1234,
"Check": tt.check,
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if resp.Code != http.StatusOK {
t.Fatalf("bad: %d", resp.Code)
}
})
t.Run(tt.name+" as slice-based service check", func(t *testing.T) {
args := map[string]interface{}{
"Name": "a",
"Port": 1234,
"Checks": []map[string]interface{}{tt.check},
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if resp.Code != http.StatusOK {
t.Fatalf("bad: %d", resp.Code)
}
})
}
}
func TestAgent_RegisterCheckScriptsExecDisable(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
args := &structs.CheckDefinition{
Name: "test",
ScriptArgs: []string{"true"},
Interval: time.Second,
}
req, _ := http.NewRequest("PUT", "/v1/agent/check/register?token=abc123", jsonReader(args))
res := httptest.NewRecorder()
a.srv.h.ServeHTTP(res, req)
if http.StatusInternalServerError != res.Code {
t.Fatalf("expected 500 code error but got %v", res.Code)
}
if !strings.Contains(res.Body.String(), "Scripts are disabled on this agent") {
t.Fatalf("expected script disabled error, got: %s", res.Body.String())
}
checkID := structs.NewCheckID("test", nil)
require.Nil(t, a.State.Check(checkID), "check registered with exec disabled")
}
func TestAgent_RegisterCheckScriptsExecRemoteDisable(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, `
enable_local_script_checks = true
`)
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
args := &structs.CheckDefinition{
Name: "test",
ScriptArgs: []string{"true"},
Interval: time.Second,
}
req, _ := http.NewRequest("PUT", "/v1/agent/check/register?token=abc123", jsonReader(args))
res := httptest.NewRecorder()
a.srv.h.ServeHTTP(res, req)
if http.StatusInternalServerError != res.Code {
t.Fatalf("expected 500 code error but got %v", res.Code)
}
if !strings.Contains(res.Body.String(), "Scripts are disabled on this agent") {
t.Fatalf("expected script disabled error, got: %s", res.Body.String())
}
checkID := structs.NewCheckID("test", nil)
require.Nil(t, a.State.Check(checkID), "check registered with exec disabled")
}
func TestAgent_RegisterCheck_Passing(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
args := &structs.CheckDefinition{
Name: "test",
TTL: 15 * time.Second,
Status: api.HealthPassing,
}
req, _ := http.NewRequest("PUT", "/v1/agent/check/register", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if http.StatusOK != resp.Code {
t.Fatalf("expcted 200 but got %v", resp.Code)
}
// Ensure we have a check mapping
checkID := structs.NewCheckID("test", nil)
if existing := a.State.Check(checkID); existing == nil {
t.Fatalf("missing test check")
}
if _, ok := a.checkTTLs[checkID]; !ok {
t.Fatalf("missing test check ttl")
}
state := a.State.Check(checkID)
if state.Status != api.HealthPassing {
t.Fatalf("bad: %v", state)
}
}
func TestAgent_RegisterCheck_BadStatus(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
args := &structs.CheckDefinition{
Name: "test",
TTL: 15 * time.Second,
Status: "fluffy",
}
req, _ := http.NewRequest("PUT", "/v1/agent/check/register", jsonReader(args))
resp := httptest.NewRecorder()
2021-11-19 16:50:44 +00:00
a.srv.h.ServeHTTP(resp, req)
if resp.Code != http.StatusBadRequest {
t.Fatalf("accepted bad status")
}
}
func TestAgent_RegisterCheck_ACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfigNew())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
nodeCheck := &structs.CheckDefinition{
Name: "test",
TTL: 15 * time.Second,
}
2017-05-09 16:58:12 +00:00
svc := &structs.ServiceDefinition{
ID: "foo:1234",
Name: "foo",
Port: 1234,
}
svcCheck := &structs.CheckDefinition{
Name: "test2",
ServiceID: "foo:1234",
TTL: 15 * time.Second,
}
// ensure the service is ready for registering a check for it.
req, _ := http.NewRequest("PUT", "/v1/agent/service/register?token=root", jsonReader(svc))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
// create a policy that has write on service foo
policyReq := &structs.ACLPolicy{
Name: "write-foo",
Rules: `service "foo" { policy = "write"}`,
}
req, _ = http.NewRequest("PUT", "/v1/acl/policy?token=root", jsonReader(policyReq))
resp = httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
// create a policy that has write on the node name of the agent
policyReq = &structs.ACLPolicy{
Name: "write-node",
Rules: fmt.Sprintf(`node "%s" { policy = "write" }`, a.config.NodeName),
}
req, _ = http.NewRequest("PUT", "/v1/acl/policy?token=root", jsonReader(policyReq))
resp = httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
// create a token using the write-foo policy
tokenReq := &structs.ACLToken{
Description: "write-foo",
Policies: []structs.ACLTokenPolicyLink{
{
Name: "write-foo",
},
},
}
req, _ = http.NewRequest("PUT", "/v1/acl/token?token=root", jsonReader(tokenReq))
resp = httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
dec := json.NewDecoder(resp.Body)
svcToken := &structs.ACLToken{}
if err := dec.Decode(svcToken); err != nil {
t.Fatalf("err: %v", err)
}
require.NotNil(t, svcToken)
// create a token using the write-node policy
tokenReq = &structs.ACLToken{
Description: "write-node",
Policies: []structs.ACLTokenPolicyLink{
{
Name: "write-node",
},
},
}
req, _ = http.NewRequest("PUT", "/v1/acl/token?token=root", jsonReader(tokenReq))
resp = httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
dec = json.NewDecoder(resp.Body)
nodeToken := &structs.ACLToken{}
if err := dec.Decode(nodeToken); err != nil {
t.Fatalf("err: %v", err)
}
require.NotNil(t, nodeToken)
t.Run("no token - node check", func(t *testing.T) {
retry.Run(t, func(r *retry.R) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/register", jsonReader(nodeCheck))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
})
t.Run("svc token - node check", func(t *testing.T) {
retry.Run(t, func(r *retry.R) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/register?token="+svcToken.SecretID, jsonReader(nodeCheck))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
})
t.Run("node token - node check", func(t *testing.T) {
retry.Run(t, func(r *retry.R) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/register?token="+nodeToken.SecretID, jsonReader(nodeCheck))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
})
t.Run("no token - svc check", func(t *testing.T) {
retry.Run(t, func(r *retry.R) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/register", jsonReader(svcCheck))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
})
t.Run("node token - svc check", func(t *testing.T) {
retry.Run(t, func(r *retry.R) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/register?token="+nodeToken.SecretID, jsonReader(svcCheck))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
})
t.Run("svc token - svc check", func(t *testing.T) {
retry.Run(t, func(r *retry.R) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/register?token="+svcToken.SecretID, jsonReader(svcCheck))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
})
}
func TestAgent_DeregisterCheck(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
chk := &structs.HealthCheck{Name: "test", CheckID: "test"}
if err := a.AddCheck(chk, nil, false, "", ConfigSourceLocal); err != nil {
t.Fatalf("err: %v", err)
}
t.Run("remove registered check", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/deregister/test", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
t.Run("remove non-existent check", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/deregister/test", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusNotFound, resp.Code)
})
// Ensure we have a check mapping
requireCheckMissing(t, a, "test")
}
func TestAgent_DeregisterCheckACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1", testrpc.WithToken("root"))
chk := &structs.HealthCheck{Name: "test", CheckID: "test"}
if err := a.AddCheck(chk, nil, false, "", ConfigSourceLocal); err != nil {
t.Fatalf("err: %v", err)
}
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/deregister/test", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("root token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/deregister/test?token=root", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
t.Run("non-existent check without token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/deregister/_nope_", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusNotFound, resp.Code)
})
t.Run("non-existent check with token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/deregister/_nope_?token=root", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusNotFound, resp.Code)
})
}
func TestAgent_PassCheck(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
chk := &structs.HealthCheck{Name: "test", CheckID: "test"}
chkType := &structs.CheckType{TTL: 15 * time.Second}
if err := a.AddCheck(chk, chkType, false, "", ConfigSourceLocal); err != nil {
t.Fatalf("err: %v", err)
}
req, _ := http.NewRequest("PUT", "/v1/agent/check/pass/test", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if http.StatusOK != resp.Code {
t.Fatalf("expected 200 by got %v", resp.Code)
}
// Ensure we have a check mapping
state := a.State.Check(structs.NewCheckID("test", nil))
if state.Status != api.HealthPassing {
t.Fatalf("bad: %v", state)
}
}
func TestAgent_PassCheck_ACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
chk := &structs.HealthCheck{Name: "test", CheckID: "test"}
chkType := &structs.CheckType{TTL: 15 * time.Second}
if err := a.AddCheck(chk, chkType, false, "", ConfigSourceLocal); err != nil {
t.Fatalf("err: %v", err)
}
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/pass/test", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("root token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/pass/test?token=root", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
}
func TestAgent_WarnCheck(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
chk := &structs.HealthCheck{Name: "test", CheckID: "test"}
chkType := &structs.CheckType{TTL: 15 * time.Second}
if err := a.AddCheck(chk, chkType, false, "", ConfigSourceLocal); err != nil {
t.Fatalf("err: %v", err)
}
req, _ := http.NewRequest("PUT", "/v1/agent/check/warn/test", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if http.StatusOK != resp.Code {
t.Fatalf("expected 200 by got %v", resp.Code)
}
// Ensure we have a check mapping
state := a.State.Check(structs.NewCheckID("test", nil))
if state.Status != api.HealthWarning {
t.Fatalf("bad: %v", state)
}
}
func TestAgent_WarnCheck_ACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
chk := &structs.HealthCheck{Name: "test", CheckID: "test"}
chkType := &structs.CheckType{TTL: 15 * time.Second}
if err := a.AddCheck(chk, chkType, false, "", ConfigSourceLocal); err != nil {
t.Fatalf("err: %v", err)
}
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/warn/test", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("root token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/warn/test?token=root", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
}
func TestAgent_FailCheck(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
chk := &structs.HealthCheck{Name: "test", CheckID: "test"}
chkType := &structs.CheckType{TTL: 15 * time.Second}
if err := a.AddCheck(chk, chkType, false, "", ConfigSourceLocal); err != nil {
t.Fatalf("err: %v", err)
}
req, _ := http.NewRequest("PUT", "/v1/agent/check/fail/test", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if http.StatusOK != resp.Code {
t.Fatalf("expected 200 by got %v", resp.Code)
}
// Ensure we have a check mapping
state := a.State.Check(structs.NewCheckID("test", nil))
if state.Status != api.HealthCritical {
t.Fatalf("bad: %v", state)
}
}
func TestAgent_FailCheck_ACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
chk := &structs.HealthCheck{Name: "test", CheckID: "test"}
chkType := &structs.CheckType{TTL: 15 * time.Second}
if err := a.AddCheck(chk, chkType, false, "", ConfigSourceLocal); err != nil {
t.Fatalf("err: %v", err)
}
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/fail/test", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("root token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/fail/test?token=root", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
}
func TestAgent_UpdateCheck(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
maxChecksSize := 256
a := NewTestAgent(t, fmt.Sprintf("check_output_max_size=%d", maxChecksSize))
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
chk := &structs.HealthCheck{Name: "test", CheckID: "test"}
chkType := &structs.CheckType{TTL: 15 * time.Second}
if err := a.AddCheck(chk, chkType, false, "", ConfigSourceLocal); err != nil {
t.Fatalf("err: %v", err)
}
cases := []checkUpdate{
{api.HealthPassing, "hello-passing"},
{api.HealthCritical, "hello-critical"},
{api.HealthWarning, "hello-warning"},
}
for _, c := range cases {
t.Run(c.Status, func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/check/update/test", jsonReader(c))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if resp.Code != http.StatusOK {
t.Fatalf("expected 200, got %d", resp.Code)
}
state := a.State.Check(structs.NewCheckID("test", nil))
if state.Status != c.Status || state.Output != c.Output {
t.Fatalf("bad: %v", state)
}
})
}
t.Run("log output limit", func(t *testing.T) {
2017-05-09 16:58:12 +00:00
args := checkUpdate{
Status: api.HealthPassing,
Output: strings.Repeat("-= bad -=", 5*maxChecksSize),
}
2017-05-09 16:58:12 +00:00
req, _ := http.NewRequest("PUT", "/v1/agent/check/update/test", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if resp.Code != http.StatusOK {
t.Fatalf("expected 200, got %d", resp.Code)
}
// Since we append some notes about truncating, we just do a
// rough check that the output buffer was cut down so this test
// isn't super brittle.
state := a.State.Check(structs.NewCheckID("test", nil))
if state.Status != api.HealthPassing || len(state.Output) > 2*maxChecksSize {
t.Fatalf("bad: %v, (len:=%d)", state, len(state.Output))
}
})
t.Run("bogus status", func(t *testing.T) {
2017-05-09 16:58:12 +00:00
args := checkUpdate{Status: "itscomplicated"}
req, _ := http.NewRequest("PUT", "/v1/agent/check/update/test", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if resp.Code != http.StatusBadRequest {
t.Fatalf("expected 400, got %d", resp.Code)
}
})
}
func TestAgent_UpdateCheck_ACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
chk := &structs.HealthCheck{Name: "test", CheckID: "test"}
chkType := &structs.CheckType{TTL: 15 * time.Second}
if err := a.AddCheck(chk, chkType, false, "", ConfigSourceLocal); err != nil {
t.Fatalf("err: %v", err)
}
t.Run("no token", func(t *testing.T) {
args := checkUpdate{api.HealthPassing, "hello-passing"}
req, _ := http.NewRequest("PUT", "/v1/agent/check/update/test", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("root token", func(t *testing.T) {
args := checkUpdate{api.HealthPassing, "hello-passing"}
req, _ := http.NewRequest("PUT", "/v1/agent/check/update/test?token=root", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
}
func TestAgent_RegisterService(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Run("normal", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService(t, "enable_central_service_config = false")
})
t.Run("service manager", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService(t, "enable_central_service_config = true")
})
}
func testAgent_RegisterService(t *testing.T, extraHCL string) {
t.Helper()
a := NewTestAgent(t, extraHCL)
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
args := &structs.ServiceDefinition{
Name: "test",
Meta: map[string]string{"hello": "world"},
Tags: []string{"primary"},
Port: 8000,
Check: structs.CheckType{
TTL: 15 * time.Second,
},
Checks: []*structs.CheckType{
{
TTL: 20 * time.Second,
},
{
TTL: 30 * time.Second,
},
},
2018-09-07 14:30:47 +00:00
Weights: &structs.Weights{
Passing: 100,
Warning: 3,
},
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register?token=abc123", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if http.StatusOK != resp.Code {
t.Fatalf("expected 200 but got %v", resp.Code)
}
// Ensure the service
sid := structs.NewServiceID("test", nil)
svc := a.State.Service(sid)
if svc == nil {
t.Fatalf("missing test service")
}
if val := svc.Meta["hello"]; val != "world" {
t.Fatalf("Missing meta: %v", svc.Meta)
}
if val := svc.Weights.Passing; val != 100 {
2018-09-07 14:30:47 +00:00
t.Fatalf("Expected 100 for Weights.Passing, got: %v", val)
}
if val := svc.Weights.Warning; val != 3 {
2018-09-07 14:30:47 +00:00
t.Fatalf("Expected 3 for Weights.Warning, got: %v", val)
}
// Ensure we have a check mapping
checks := a.State.Checks(structs.WildcardEnterpriseMetaInDefaultPartition())
if len(checks) != 3 {
t.Fatalf("bad: %v", checks)
}
2019-10-17 18:33:11 +00:00
for _, c := range checks {
if c.Type != "ttl" {
t.Fatalf("expected ttl check type, got %s", c.Type)
}
}
if len(a.checkTTLs) != 3 {
t.Fatalf("missing test check ttls: %v", a.checkTTLs)
}
2015-04-28 20:06:02 +00:00
// Ensure the token was configured
if token := a.State.ServiceToken(sid); token == "" {
2015-04-28 20:06:02 +00:00
t.Fatalf("missing token")
}
}
func TestAgent_RegisterService_ReRegister(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Run("normal", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_ReRegister(t, "enable_central_service_config = false")
})
t.Run("service manager", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_ReRegister(t, "enable_central_service_config = true")
})
}
func testAgent_RegisterService_ReRegister(t *testing.T, extraHCL string) {
t.Helper()
a := NewTestAgent(t, extraHCL)
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
args := &structs.ServiceDefinition{
Name: "test",
Meta: map[string]string{"hello": "world"},
Tags: []string{"primary"},
Port: 8000,
Checks: []*structs.CheckType{
{
CheckID: types.CheckID("check_1"),
TTL: 20 * time.Second,
},
{
CheckID: types.CheckID("check_2"),
TTL: 30 * time.Second,
},
},
Weights: &structs.Weights{
Passing: 100,
Warning: 3,
},
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
args = &structs.ServiceDefinition{
Name: "test",
Meta: map[string]string{"hello": "world"},
Tags: []string{"primary"},
Port: 8000,
Checks: []*structs.CheckType{
{
CheckID: types.CheckID("check_1"),
TTL: 20 * time.Second,
},
{
CheckID: types.CheckID("check_3"),
TTL: 30 * time.Second,
},
},
Weights: &structs.Weights{
Passing: 100,
Warning: 3,
},
}
req, _ = http.NewRequest("PUT", "/v1/agent/service/register", jsonReader(args))
resp = httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
checks := a.State.Checks(structs.DefaultEnterpriseMetaInDefaultPartition())
require.Equal(t, 3, len(checks))
checkIDs := []string{}
for id := range checks {
checkIDs = append(checkIDs, string(id.ID))
}
require.ElementsMatch(t, []string{"check_1", "check_2", "check_3"}, checkIDs)
}
func TestAgent_RegisterService_ReRegister_ReplaceExistingChecks(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Run("normal", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_ReRegister_ReplaceExistingChecks(t, "enable_central_service_config = false")
})
t.Run("service manager", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_ReRegister_ReplaceExistingChecks(t, "enable_central_service_config = true")
})
}
func testAgent_RegisterService_ReRegister_ReplaceExistingChecks(t *testing.T, extraHCL string) {
t.Helper()
a := NewTestAgent(t, extraHCL)
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
args := &structs.ServiceDefinition{
Name: "test",
Meta: map[string]string{"hello": "world"},
Tags: []string{"primary"},
Port: 8000,
Checks: []*structs.CheckType{
{
// explicitly not setting the check id to let it be auto-generated
2019-11-15 15:33:21 +00:00
// we want to ensure that we are testing out the cases with autogenerated names/ids
TTL: 20 * time.Second,
},
{
CheckID: types.CheckID("check_2"),
TTL: 30 * time.Second,
},
},
Weights: &structs.Weights{
Passing: 100,
Warning: 3,
},
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register?replace-existing-checks", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
args = &structs.ServiceDefinition{
Name: "test",
Meta: map[string]string{"hello": "world"},
Tags: []string{"primary"},
Port: 8000,
Checks: []*structs.CheckType{
{
TTL: 20 * time.Second,
},
{
CheckID: types.CheckID("check_3"),
TTL: 30 * time.Second,
},
},
Weights: &structs.Weights{
Passing: 100,
Warning: 3,
},
}
req, _ = http.NewRequest("PUT", "/v1/agent/service/register?replace-existing-checks", jsonReader(args))
resp = httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
checks := a.State.Checks(structs.DefaultEnterpriseMetaInDefaultPartition())
require.Len(t, checks, 2)
checkIDs := []string{}
for id := range checks {
checkIDs = append(checkIDs, string(id.ID))
}
require.ElementsMatch(t, []string{"service:test:1", "check_3"}, checkIDs)
}
func TestAgent_RegisterService_TranslateKeys(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Run("normal", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_TranslateKeys(t, "enable_central_service_config = false")
})
t.Run("service manager", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_TranslateKeys(t, "enable_central_service_config = true")
})
}
func testAgent_RegisterService_TranslateKeys(t *testing.T, extraHCL string) {
t.Helper()
tests := []struct {
ip string
expectedTCPCheckStart string
}{
{"127.0.0.1", "127.0.0.1:"}, // private network address
{"::1", "[::1]:"}, // shared address space
}
for _, tt := range tests {
t.Run(tt.ip, func(t *testing.T) {
a := NewTestAgent(t, `
connect {}
`+extraHCL)
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
json := `
Add Proxy Upstreams to Service Definition (#4639) * Refactor Service Definition ProxyDestination. This includes: - Refactoring all internal structs used - Updated tests for both deprecated and new input for: - Agent Services endpoint response - Agent Service endpoint response - Agent Register endpoint - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Register - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Services endpoint response - Catalog Node endpoint response - Catalog Service endpoint response - Updated API tests for all of the above too (both deprecated and new forms of register) TODO: - config package changes for on-disk service definitions - proxy config endpoint - built-in proxy support for new fields * Agent proxy config endpoint updated with upstreams * Config file changes for upstreams. * Add upstream opaque config and update all tests to ensure it works everywhere. * Built in proxy working with new Upstreams config * Command fixes and deprecations * Fix key translation, upstream type defaults and a spate of other subtele bugs found with ned to end test scripts... TODO: tests still failing on one case that needs a fix. I think it's key translation for upstreams nested in Managed proxy struct. * Fix translated keys in API registration. ≈ * Fixes from docs - omit some empty undocumented fields in API - Bring back ServiceProxyDestination in Catalog responses to not break backwards compat - this was removed assuming it was only used internally. * Documentation updates for Upstreams in service definition * Fixes for tests broken by many refactors. * Enable travis on f-connect branch in this branch too. * Add consistent Deprecation comments to ProxyDestination uses * Update version number on deprecation notices, and correct upstream datacenter field with explanation in docs
2018-09-12 16:07:47 +00:00
{
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
"name":"test",
"port":8000,
"enable_tag_override": true,
"tagged_addresses": {
"lan": {
"address": "1.2.3.4",
"port": 5353
},
"wan": {
"address": "2.3.4.5",
"port": 53
}
},
Add Proxy Upstreams to Service Definition (#4639) * Refactor Service Definition ProxyDestination. This includes: - Refactoring all internal structs used - Updated tests for both deprecated and new input for: - Agent Services endpoint response - Agent Service endpoint response - Agent Register endpoint - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Register - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Services endpoint response - Catalog Node endpoint response - Catalog Service endpoint response - Updated API tests for all of the above too (both deprecated and new forms of register) TODO: - config package changes for on-disk service definitions - proxy config endpoint - built-in proxy support for new fields * Agent proxy config endpoint updated with upstreams * Config file changes for upstreams. * Add upstream opaque config and update all tests to ensure it works everywhere. * Built in proxy working with new Upstreams config * Command fixes and deprecations * Fix key translation, upstream type defaults and a spate of other subtele bugs found with ned to end test scripts... TODO: tests still failing on one case that needs a fix. I think it's key translation for upstreams nested in Managed proxy struct. * Fix translated keys in API registration. ≈ * Fixes from docs - omit some empty undocumented fields in API - Bring back ServiceProxyDestination in Catalog responses to not break backwards compat - this was removed assuming it was only used internally. * Documentation updates for Upstreams in service definition * Fixes for tests broken by many refactors. * Enable travis on f-connect branch in this branch too. * Add consistent Deprecation comments to ProxyDestination uses * Update version number on deprecation notices, and correct upstream datacenter field with explanation in docs
2018-09-12 16:07:47 +00:00
"meta": {
"some": "meta",
"enable_tag_override": "meta is 'opaque' so should not get translated"
Add Proxy Upstreams to Service Definition (#4639) * Refactor Service Definition ProxyDestination. This includes: - Refactoring all internal structs used - Updated tests for both deprecated and new input for: - Agent Services endpoint response - Agent Service endpoint response - Agent Register endpoint - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Register - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Services endpoint response - Catalog Node endpoint response - Catalog Service endpoint response - Updated API tests for all of the above too (both deprecated and new forms of register) TODO: - config package changes for on-disk service definitions - proxy config endpoint - built-in proxy support for new fields * Agent proxy config endpoint updated with upstreams * Config file changes for upstreams. * Add upstream opaque config and update all tests to ensure it works everywhere. * Built in proxy working with new Upstreams config * Command fixes and deprecations * Fix key translation, upstream type defaults and a spate of other subtele bugs found with ned to end test scripts... TODO: tests still failing on one case that needs a fix. I think it's key translation for upstreams nested in Managed proxy struct. * Fix translated keys in API registration. ≈ * Fixes from docs - omit some empty undocumented fields in API - Bring back ServiceProxyDestination in Catalog responses to not break backwards compat - this was removed assuming it was only used internally. * Documentation updates for Upstreams in service definition * Fixes for tests broken by many refactors. * Enable travis on f-connect branch in this branch too. * Add consistent Deprecation comments to ProxyDestination uses * Update version number on deprecation notices, and correct upstream datacenter field with explanation in docs
2018-09-12 16:07:47 +00:00
},
"kind": "connect-proxy",` +
// Note the uppercase P is important here - it ensures translation works
// correctly in case-insensitive way. Without it this test can pass even
// when translation is broken for other valid inputs.
`"Proxy": {
Add Proxy Upstreams to Service Definition (#4639) * Refactor Service Definition ProxyDestination. This includes: - Refactoring all internal structs used - Updated tests for both deprecated and new input for: - Agent Services endpoint response - Agent Service endpoint response - Agent Register endpoint - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Register - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Services endpoint response - Catalog Node endpoint response - Catalog Service endpoint response - Updated API tests for all of the above too (both deprecated and new forms of register) TODO: - config package changes for on-disk service definitions - proxy config endpoint - built-in proxy support for new fields * Agent proxy config endpoint updated with upstreams * Config file changes for upstreams. * Add upstream opaque config and update all tests to ensure it works everywhere. * Built in proxy working with new Upstreams config * Command fixes and deprecations * Fix key translation, upstream type defaults and a spate of other subtele bugs found with ned to end test scripts... TODO: tests still failing on one case that needs a fix. I think it's key translation for upstreams nested in Managed proxy struct. * Fix translated keys in API registration. ≈ * Fixes from docs - omit some empty undocumented fields in API - Bring back ServiceProxyDestination in Catalog responses to not break backwards compat - this was removed assuming it was only used internally. * Documentation updates for Upstreams in service definition * Fixes for tests broken by many refactors. * Enable travis on f-connect branch in this branch too. * Add consistent Deprecation comments to ProxyDestination uses * Update version number on deprecation notices, and correct upstream datacenter field with explanation in docs
2018-09-12 16:07:47 +00:00
"destination_service_name": "web",
"destination_service_id": "web",
"local_service_port": 1234,
"local_service_address": "` + tt.ip + `",
"config": {
"destination_type": "proxy.config is 'opaque' so should not get translated"
},
Add Proxy Upstreams to Service Definition (#4639) * Refactor Service Definition ProxyDestination. This includes: - Refactoring all internal structs used - Updated tests for both deprecated and new input for: - Agent Services endpoint response - Agent Service endpoint response - Agent Register endpoint - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Register - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Services endpoint response - Catalog Node endpoint response - Catalog Service endpoint response - Updated API tests for all of the above too (both deprecated and new forms of register) TODO: - config package changes for on-disk service definitions - proxy config endpoint - built-in proxy support for new fields * Agent proxy config endpoint updated with upstreams * Config file changes for upstreams. * Add upstream opaque config and update all tests to ensure it works everywhere. * Built in proxy working with new Upstreams config * Command fixes and deprecations * Fix key translation, upstream type defaults and a spate of other subtele bugs found with ned to end test scripts... TODO: tests still failing on one case that needs a fix. I think it's key translation for upstreams nested in Managed proxy struct. * Fix translated keys in API registration. ≈ * Fixes from docs - omit some empty undocumented fields in API - Bring back ServiceProxyDestination in Catalog responses to not break backwards compat - this was removed assuming it was only used internally. * Documentation updates for Upstreams in service definition * Fixes for tests broken by many refactors. * Enable travis on f-connect branch in this branch too. * Add consistent Deprecation comments to ProxyDestination uses * Update version number on deprecation notices, and correct upstream datacenter field with explanation in docs
2018-09-12 16:07:47 +00:00
"upstreams": [
{
"destination_type": "service",
"destination_namespace": "default",
2021-08-20 16:57:45 +00:00
"destination_partition": "default",
Add Proxy Upstreams to Service Definition (#4639) * Refactor Service Definition ProxyDestination. This includes: - Refactoring all internal structs used - Updated tests for both deprecated and new input for: - Agent Services endpoint response - Agent Service endpoint response - Agent Register endpoint - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Register - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Services endpoint response - Catalog Node endpoint response - Catalog Service endpoint response - Updated API tests for all of the above too (both deprecated and new forms of register) TODO: - config package changes for on-disk service definitions - proxy config endpoint - built-in proxy support for new fields * Agent proxy config endpoint updated with upstreams * Config file changes for upstreams. * Add upstream opaque config and update all tests to ensure it works everywhere. * Built in proxy working with new Upstreams config * Command fixes and deprecations * Fix key translation, upstream type defaults and a spate of other subtele bugs found with ned to end test scripts... TODO: tests still failing on one case that needs a fix. I think it's key translation for upstreams nested in Managed proxy struct. * Fix translated keys in API registration. ≈ * Fixes from docs - omit some empty undocumented fields in API - Bring back ServiceProxyDestination in Catalog responses to not break backwards compat - this was removed assuming it was only used internally. * Documentation updates for Upstreams in service definition * Fixes for tests broken by many refactors. * Enable travis on f-connect branch in this branch too. * Add consistent Deprecation comments to ProxyDestination uses * Update version number on deprecation notices, and correct upstream datacenter field with explanation in docs
2018-09-12 16:07:47 +00:00
"destination_name": "db",
"local_bind_address": "` + tt.ip + `",
"local_bind_port": 1234,
"config": {
"destination_type": "proxy.upstreams.config is 'opaque' so should not get translated"
}
Add Proxy Upstreams to Service Definition (#4639) * Refactor Service Definition ProxyDestination. This includes: - Refactoring all internal structs used - Updated tests for both deprecated and new input for: - Agent Services endpoint response - Agent Service endpoint response - Agent Register endpoint - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Register - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Services endpoint response - Catalog Node endpoint response - Catalog Service endpoint response - Updated API tests for all of the above too (both deprecated and new forms of register) TODO: - config package changes for on-disk service definitions - proxy config endpoint - built-in proxy support for new fields * Agent proxy config endpoint updated with upstreams * Config file changes for upstreams. * Add upstream opaque config and update all tests to ensure it works everywhere. * Built in proxy working with new Upstreams config * Command fixes and deprecations * Fix key translation, upstream type defaults and a spate of other subtele bugs found with ned to end test scripts... TODO: tests still failing on one case that needs a fix. I think it's key translation for upstreams nested in Managed proxy struct. * Fix translated keys in API registration. ≈ * Fixes from docs - omit some empty undocumented fields in API - Bring back ServiceProxyDestination in Catalog responses to not break backwards compat - this was removed assuming it was only used internally. * Documentation updates for Upstreams in service definition * Fixes for tests broken by many refactors. * Enable travis on f-connect branch in this branch too. * Add consistent Deprecation comments to ProxyDestination uses * Update version number on deprecation notices, and correct upstream datacenter field with explanation in docs
2018-09-12 16:07:47 +00:00
}
]
},
"connect": {
"sidecar_service": {
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
"name":"test-proxy",
"port":8001,
"enable_tag_override": true,
"meta": {
"some": "meta",
"enable_tag_override": "sidecar_service.meta is 'opaque' so should not get translated"
},
"kind": "connect-proxy",
"proxy": {
"destination_service_name": "test",
"destination_service_id": "test",
"local_service_port": 4321,
"local_service_address": "` + tt.ip + `",
"upstreams": [
{
"destination_type": "service",
"destination_namespace": "default",
2021-08-20 16:57:45 +00:00
"destination_partition": "default",
"destination_name": "db",
"local_bind_address": "` + tt.ip + `",
"local_bind_port": 1234,
"config": {
"destination_type": "sidecar_service.proxy.upstreams.config is 'opaque' so should not get translated"
}
}
]
}
Add Proxy Upstreams to Service Definition (#4639) * Refactor Service Definition ProxyDestination. This includes: - Refactoring all internal structs used - Updated tests for both deprecated and new input for: - Agent Services endpoint response - Agent Service endpoint response - Agent Register endpoint - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Register - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Services endpoint response - Catalog Node endpoint response - Catalog Service endpoint response - Updated API tests for all of the above too (both deprecated and new forms of register) TODO: - config package changes for on-disk service definitions - proxy config endpoint - built-in proxy support for new fields * Agent proxy config endpoint updated with upstreams * Config file changes for upstreams. * Add upstream opaque config and update all tests to ensure it works everywhere. * Built in proxy working with new Upstreams config * Command fixes and deprecations * Fix key translation, upstream type defaults and a spate of other subtele bugs found with ned to end test scripts... TODO: tests still failing on one case that needs a fix. I think it's key translation for upstreams nested in Managed proxy struct. * Fix translated keys in API registration. ≈ * Fixes from docs - omit some empty undocumented fields in API - Bring back ServiceProxyDestination in Catalog responses to not break backwards compat - this was removed assuming it was only used internally. * Documentation updates for Upstreams in service definition * Fixes for tests broken by many refactors. * Enable travis on f-connect branch in this branch too. * Add consistent Deprecation comments to ProxyDestination uses * Update version number on deprecation notices, and correct upstream datacenter field with explanation in docs
2018-09-12 16:07:47 +00:00
}
},
"weights":{
"passing": 16
}
}`
req, _ := http.NewRequest("PUT", "/v1/agent/service/register", strings.NewReader(json))
rr := httptest.NewRecorder()
a.srv.h.ServeHTTP(rr, req)
require.Equal(t, 200, rr.Code, "body: %s", rr.Body)
svc := &structs.NodeService{
ID: "test",
Service: "test",
TaggedAddresses: map[string]structs.ServiceAddress{
"lan": {
Address: "1.2.3.4",
Port: 5353,
},
"wan": {
Address: "2.3.4.5",
Port: 53,
},
},
Meta: map[string]string{
"some": "meta",
"enable_tag_override": "meta is 'opaque' so should not get translated",
},
Port: 8000,
EnableTagOverride: true,
Weights: &structs.Weights{Passing: 16, Warning: 0},
Kind: structs.ServiceKindConnectProxy,
Proxy: structs.ConnectProxyConfig{
DestinationServiceName: "web",
DestinationServiceID: "web",
LocalServiceAddress: tt.ip,
LocalServicePort: 1234,
Config: map[string]interface{}{
"destination_type": "proxy.config is 'opaque' so should not get translated",
},
Upstreams: structs.Upstreams{
{
DestinationType: structs.UpstreamDestTypeService,
DestinationName: "db",
DestinationNamespace: "default",
2021-08-20 16:57:45 +00:00
DestinationPartition: "default",
LocalBindAddress: tt.ip,
LocalBindPort: 1234,
Config: map[string]interface{}{
"destination_type": "proxy.upstreams.config is 'opaque' so should not get translated",
},
},
},
Add Proxy Upstreams to Service Definition (#4639) * Refactor Service Definition ProxyDestination. This includes: - Refactoring all internal structs used - Updated tests for both deprecated and new input for: - Agent Services endpoint response - Agent Service endpoint response - Agent Register endpoint - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Register - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Services endpoint response - Catalog Node endpoint response - Catalog Service endpoint response - Updated API tests for all of the above too (both deprecated and new forms of register) TODO: - config package changes for on-disk service definitions - proxy config endpoint - built-in proxy support for new fields * Agent proxy config endpoint updated with upstreams * Config file changes for upstreams. * Add upstream opaque config and update all tests to ensure it works everywhere. * Built in proxy working with new Upstreams config * Command fixes and deprecations * Fix key translation, upstream type defaults and a spate of other subtele bugs found with ned to end test scripts... TODO: tests still failing on one case that needs a fix. I think it's key translation for upstreams nested in Managed proxy struct. * Fix translated keys in API registration. ≈ * Fixes from docs - omit some empty undocumented fields in API - Bring back ServiceProxyDestination in Catalog responses to not break backwards compat - this was removed assuming it was only used internally. * Documentation updates for Upstreams in service definition * Fixes for tests broken by many refactors. * Enable travis on f-connect branch in this branch too. * Add consistent Deprecation comments to ProxyDestination uses * Update version number on deprecation notices, and correct upstream datacenter field with explanation in docs
2018-09-12 16:07:47 +00:00
},
Connect: structs.ServiceConnect{
// The sidecar service is nilled since it is only config sugar and
// shouldn't be represented in state. We assert that the translations
// there worked by inspecting the registered sidecar below.
SidecarService: nil,
},
EnterpriseMeta: *structs.DefaultEnterpriseMetaInDefaultPartition(),
}
got := a.State.Service(structs.NewServiceID("test", nil))
require.Equal(t, svc, got)
sidecarSvc := &structs.NodeService{
Kind: structs.ServiceKindConnectProxy,
ID: "test-sidecar-proxy",
Service: "test-proxy",
Meta: map[string]string{
"some": "meta",
"enable_tag_override": "sidecar_service.meta is 'opaque' so should not get translated",
},
TaggedAddresses: map[string]structs.ServiceAddress{},
Port: 8001,
EnableTagOverride: true,
Weights: &structs.Weights{Passing: 1, Warning: 1},
LocallyRegisteredAsSidecar: true,
Proxy: structs.ConnectProxyConfig{
DestinationServiceName: "test",
DestinationServiceID: "test",
LocalServiceAddress: tt.ip,
LocalServicePort: 4321,
Upstreams: structs.Upstreams{
{
DestinationType: structs.UpstreamDestTypeService,
DestinationName: "db",
DestinationNamespace: "default",
2021-08-20 16:57:45 +00:00
DestinationPartition: "default",
LocalBindAddress: tt.ip,
LocalBindPort: 1234,
Config: map[string]interface{}{
"destination_type": "sidecar_service.proxy.upstreams.config is 'opaque' so should not get translated",
},
},
Add Proxy Upstreams to Service Definition (#4639) * Refactor Service Definition ProxyDestination. This includes: - Refactoring all internal structs used - Updated tests for both deprecated and new input for: - Agent Services endpoint response - Agent Service endpoint response - Agent Register endpoint - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Register - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Services endpoint response - Catalog Node endpoint response - Catalog Service endpoint response - Updated API tests for all of the above too (both deprecated and new forms of register) TODO: - config package changes for on-disk service definitions - proxy config endpoint - built-in proxy support for new fields * Agent proxy config endpoint updated with upstreams * Config file changes for upstreams. * Add upstream opaque config and update all tests to ensure it works everywhere. * Built in proxy working with new Upstreams config * Command fixes and deprecations * Fix key translation, upstream type defaults and a spate of other subtele bugs found with ned to end test scripts... TODO: tests still failing on one case that needs a fix. I think it's key translation for upstreams nested in Managed proxy struct. * Fix translated keys in API registration. ≈ * Fixes from docs - omit some empty undocumented fields in API - Bring back ServiceProxyDestination in Catalog responses to not break backwards compat - this was removed assuming it was only used internally. * Documentation updates for Upstreams in service definition * Fixes for tests broken by many refactors. * Enable travis on f-connect branch in this branch too. * Add consistent Deprecation comments to ProxyDestination uses * Update version number on deprecation notices, and correct upstream datacenter field with explanation in docs
2018-09-12 16:07:47 +00:00
},
},
EnterpriseMeta: *structs.DefaultEnterpriseMetaInDefaultPartition(),
}
gotSidecar := a.State.Service(structs.NewServiceID("test-sidecar-proxy", nil))
hasNoCorrectTCPCheck := true
for _, v := range a.checkTCPs {
if strings.HasPrefix(v.TCP, tt.expectedTCPCheckStart) {
hasNoCorrectTCPCheck = false
break
}
fmt.Println("TCP Check:= ", v)
}
if hasNoCorrectTCPCheck {
t.Fatalf("Did not find the expected TCP Healtcheck '%s' in %#v ", tt.expectedTCPCheckStart, a.checkTCPs)
}
require.Equal(t, sidecarSvc, gotSidecar)
})
}
}
func TestAgent_RegisterService_ACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Run("normal", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_ACLDeny(t, "enable_central_service_config = false")
})
t.Run("service manager", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_ACLDeny(t, "enable_central_service_config = true")
})
}
func testAgent_RegisterService_ACLDeny(t *testing.T, extraHCL string) {
t.Helper()
a := NewTestAgent(t, TestACLConfig()+" "+extraHCL)
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
args := &structs.ServiceDefinition{
Name: "test",
Tags: []string{"primary"},
Port: 8000,
Check: structs.CheckType{
TTL: 15 * time.Second,
},
Checks: []*structs.CheckType{
{
TTL: 20 * time.Second,
},
{
TTL: 30 * time.Second,
},
},
}
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/service/register", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("root token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/service/register?token=root", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
}
func TestAgent_RegisterService_InvalidAddress(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Run("normal", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_InvalidAddress(t, "enable_central_service_config = false")
})
t.Run("service manager", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_InvalidAddress(t, "enable_central_service_config = true")
})
}
func testAgent_RegisterService_InvalidAddress(t *testing.T, extraHCL string) {
t.Helper()
a := NewTestAgent(t, extraHCL)
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
for _, addr := range []string{"0.0.0.0", "::", "[::]"} {
t.Run("addr "+addr, func(t *testing.T) {
args := &structs.ServiceDefinition{
Name: "test",
Address: addr,
Port: 8000,
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register?token=abc123", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if got, want := resp.Code, 400; got != want {
t.Fatalf("got code %d want %d", got, want)
}
})
}
}
// This tests local agent service registration of a unmanaged connect proxy.
// This verifies that it is put in the local state store properly for syncing
// later.
func TestAgent_RegisterService_UnmanagedConnectProxy(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Run("normal", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_UnmanagedConnectProxy(t, "enable_central_service_config = false")
})
t.Run("service manager", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_UnmanagedConnectProxy(t, "enable_central_service_config = true")
})
}
func testAgent_RegisterService_UnmanagedConnectProxy(t *testing.T, extraHCL string) {
t.Helper()
a := NewTestAgent(t, extraHCL)
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
Add Proxy Upstreams to Service Definition (#4639) * Refactor Service Definition ProxyDestination. This includes: - Refactoring all internal structs used - Updated tests for both deprecated and new input for: - Agent Services endpoint response - Agent Service endpoint response - Agent Register endpoint - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Register - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Services endpoint response - Catalog Node endpoint response - Catalog Service endpoint response - Updated API tests for all of the above too (both deprecated and new forms of register) TODO: - config package changes for on-disk service definitions - proxy config endpoint - built-in proxy support for new fields * Agent proxy config endpoint updated with upstreams * Config file changes for upstreams. * Add upstream opaque config and update all tests to ensure it works everywhere. * Built in proxy working with new Upstreams config * Command fixes and deprecations * Fix key translation, upstream type defaults and a spate of other subtele bugs found with ned to end test scripts... TODO: tests still failing on one case that needs a fix. I think it's key translation for upstreams nested in Managed proxy struct. * Fix translated keys in API registration. ≈ * Fixes from docs - omit some empty undocumented fields in API - Bring back ServiceProxyDestination in Catalog responses to not break backwards compat - this was removed assuming it was only used internally. * Documentation updates for Upstreams in service definition * Fixes for tests broken by many refactors. * Enable travis on f-connect branch in this branch too. * Add consistent Deprecation comments to ProxyDestination uses * Update version number on deprecation notices, and correct upstream datacenter field with explanation in docs
2018-09-12 16:07:47 +00:00
// Register a proxy. Note that the destination doesn't exist here on this
// agent or in the catalog at all. This is intended and part of the design.
args := &api.AgentServiceRegistration{
Kind: api.ServiceKindConnectProxy,
Name: "connect-proxy",
Port: 8000,
Proxy: &api.AgentServiceConnectProxyConfig{
DestinationServiceName: "web",
Upstreams: []api.Upstream{
{
// No type to force default
DestinationName: "db",
LocalBindPort: 1234,
},
{
DestinationType: "prepared_query",
DestinationName: "geo-cache",
LocalBindPort: 1235,
},
},
2021-04-13 22:08:41 +00:00
Mode: api.ProxyModeTransparent,
TransparentProxy: &api.TransparentProxyConfig{
OutboundListenerPort: 808,
},
},
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register?token=abc123", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
// Ensure the service
sid := structs.NewServiceID("connect-proxy", nil)
svc := a.State.Service(sid)
require.NotNil(t, svc, "has service")
require.Equal(t, structs.ServiceKindConnectProxy, svc.Kind)
// Registration sets default types and namespaces
for i := range args.Proxy.Upstreams {
if args.Proxy.Upstreams[i].DestinationType == "" {
args.Proxy.Upstreams[i].DestinationType = api.UpstreamDestTypeService
}
if args.Proxy.Upstreams[i].DestinationNamespace == "" {
args.Proxy.Upstreams[i].DestinationNamespace =
structs.DefaultEnterpriseMetaInDefaultPartition().NamespaceOrEmpty()
}
2021-08-20 16:57:45 +00:00
if args.Proxy.Upstreams[i].DestinationPartition == "" {
args.Proxy.Upstreams[i].DestinationPartition =
structs.DefaultEnterpriseMetaInDefaultPartition().PartitionOrEmpty()
}
}
require.Equal(t, args.Proxy, svc.Proxy.ToAPI())
// Ensure the token was configured
require.Equal(t, "abc123", a.State.ServiceToken(structs.NewServiceID("connect-proxy", nil)))
}
func testDefaultSidecar(svc string, port int, fns ...func(*structs.NodeService)) *structs.NodeService {
ns := &structs.NodeService{
ID: svc + "-sidecar-proxy",
Kind: structs.ServiceKindConnectProxy,
Service: svc + "-sidecar-proxy",
Port: 2222,
TaggedAddresses: map[string]structs.ServiceAddress{},
Weights: &structs.Weights{
Passing: 1,
Warning: 1,
},
// Note that LocallyRegisteredAsSidecar should be true on the internal
// NodeService, but that we never want to see it in the HTTP response as
// it's internal only state. This is being compared directly to local state
// so should be present here.
LocallyRegisteredAsSidecar: true,
Proxy: structs.ConnectProxyConfig{
DestinationServiceName: svc,
DestinationServiceID: svc,
LocalServiceAddress: "127.0.0.1",
LocalServicePort: port,
},
EnterpriseMeta: *structs.DefaultEnterpriseMetaInDefaultPartition(),
}
for _, fn := range fns {
fn(ns)
}
return ns
}
// testCreateToken creates a Policy for the provided rules and a Token linked to that Policy.
func testCreateToken(t *testing.T, a *TestAgent, rules string) string {
policyName, err := uuid.GenerateUUID() // we just need a unique name for the test and UUIDs are definitely unique
require.NoError(t, err)
policyID := testCreatePolicy(t, a, policyName, rules)
args := map[string]interface{}{
"Description": "User Token",
"Policies": []map[string]interface{}{
{
"ID": policyID,
},
},
"Local": false,
}
req, _ := http.NewRequest("PUT", "/v1/acl/token?token=root", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
dec := json.NewDecoder(resp.Body)
aclResp := &structs.ACLToken{}
if err := dec.Decode(aclResp); err != nil {
t.Fatalf("err: %v", err)
}
return aclResp.SecretID
}
func testCreatePolicy(t *testing.T, a *TestAgent, name, rules string) string {
args := map[string]interface{}{
"Name": name,
"Rules": rules,
}
req, _ := http.NewRequest("PUT", "/v1/acl/policy?token=root", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
dec := json.NewDecoder(resp.Body)
aclResp := &structs.ACLPolicy{}
if err := dec.Decode(aclResp); err != nil {
t.Fatalf("err: %v", err)
}
return aclResp.ID
}
// This tests local agent service registration with a sidecar service. Note we
// only test simple defaults for the sidecar here since the actual logic for
// handling sidecar defaults and port assignment is tested thoroughly in
// TestAgent_sidecarServiceFromNodeService. Note it also tests Deregister
// explicitly too since setup is identical.
func TestAgent_RegisterServiceDeregisterService_Sidecar(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Run("normal", func(t *testing.T) {
t.Parallel()
testAgent_RegisterServiceDeregisterService_Sidecar(t, "enable_central_service_config = false")
})
t.Run("service manager", func(t *testing.T) {
t.Parallel()
testAgent_RegisterServiceDeregisterService_Sidecar(t, "enable_central_service_config = true")
})
}
func testAgent_RegisterServiceDeregisterService_Sidecar(t *testing.T, extraHCL string) {
t.Helper()
tests := []struct {
name string
preRegister, preRegister2 *structs.NodeService
// Use raw JSON payloads rather than encoding to avoid subtleties with some
// internal representations and different ways they encode and decode. We
// rely on the payload being Unmarshalable to structs.ServiceDefinition
// directly.
json string
enableACL bool
tokenRules string
wantNS *structs.NodeService
wantErr string
wantSidecarIDLeftAfterDereg bool
assertStateFn func(t *testing.T, state *local.State)
}{
{
name: "sanity check no sidecar case",
json: `
{
"name": "web",
"port": 1111
}
`,
wantNS: nil,
wantErr: "",
},
{
name: "default sidecar",
json: `
{
"name": "web",
"port": 1111,
"connect": {
"SidecarService": {}
}
}
`,
wantNS: testDefaultSidecar("web", 1111),
wantErr: "",
},
{
name: "ACL OK defaults",
json: `
{
"name": "web",
"port": 1111,
"connect": {
"SidecarService": {}
}
}
`,
enableACL: true,
tokenRules: `
service "web-sidecar-proxy" {
policy = "write"
}
service "web" {
policy = "write"
}`,
wantNS: testDefaultSidecar("web", 1111),
wantErr: "",
},
{
name: "ACL denied",
json: `
{
"name": "web",
"port": 1111,
"connect": {
"SidecarService": {}
}
}
`,
enableACL: true,
tokenRules: ``, // No token rules means no valid token
wantNS: nil,
wantErr: "Permission denied",
},
{
name: "ACL OK for service but not for sidecar",
json: `
{
"name": "web",
"port": 1111,
"connect": {
"SidecarService": {}
}
}
`,
enableACL: true,
// This will become more common/reasonable when ACLs support exact match.
tokenRules: `
service "web-sidecar-proxy" {
policy = "deny"
}
service "web" {
policy = "write"
}`,
wantNS: nil,
wantErr: "Permission denied",
},
{
name: "ACL OK for service and sidecar but not sidecar's overridden destination",
json: `
{
"name": "web",
"port": 1111,
"connect": {
"SidecarService": {
"proxy": {
"DestinationServiceName": "foo"
}
}
}
}
`,
enableACL: true,
tokenRules: `
service "web-sidecar-proxy" {
policy = "write"
}
service "web" {
policy = "write"
}`,
wantNS: nil,
wantErr: "Permission denied",
},
{
name: "ACL OK for service but not for overridden sidecar",
json: `
{
"name": "web",
"port": 1111,
"connect": {
"SidecarService": {
"name": "foo-sidecar-proxy"
}
}
}
`,
enableACL: true,
tokenRules: `
service "web-sidecar-proxy" {
policy = "write"
}
service "web" {
policy = "write"
}`,
wantNS: nil,
wantErr: "Permission denied",
},
{
name: "ACL OK for service but and overridden for sidecar",
// This test ensures that if the sidecar embeds it's own token with
// different privs from the main request token it will be honored for the
// sidecar registration. We use the test root token since that should have
// permission.
json: `
{
"name": "web",
"port": 1111,
"connect": {
"SidecarService": {
"name": "foo",
"token": "root"
}
}
}
`,
enableACL: true,
tokenRules: `
service "web-sidecar-proxy" {
policy = "write"
}
service "web" {
policy = "write"
}`,
wantNS: testDefaultSidecar("web", 1111, func(ns *structs.NodeService) {
ns.Service = "foo"
}),
wantErr: "",
},
{
name: "invalid check definition in sidecar",
// Note no interval in the TCP check should fail validation
json: `
{
"name": "web",
"port": 1111,
"connect": {
"SidecarService": {
"check": {
"TCP": "foo"
}
}
}
}
`,
wantNS: nil,
wantErr: "invalid check in sidecar_service",
},
{
name: "invalid checks definitions in sidecar",
// Note no interval in the TCP check should fail validation
json: `
{
"name": "web",
"port": 1111,
"connect": {
"SidecarService": {
"checks": [{
"TCP": "foo"
}]
}
}
}
`,
wantNS: nil,
wantErr: "invalid check in sidecar_service",
},
{
name: "invalid check status in sidecar",
// Note no interval in the TCP check should fail validation
json: `
{
"name": "web",
"port": 1111,
"connect": {
"SidecarService": {
"check": {
"TCP": "foo",
"Interval": 10,
"Status": "unsupported-status"
}
}
}
}
`,
wantNS: nil,
wantErr: "Status for checks must 'passing', 'warning', 'critical'",
},
{
name: "invalid checks status in sidecar",
// Note no interval in the TCP check should fail validation
json: `
{
"name": "web",
"port": 1111,
"connect": {
"SidecarService": {
"checks": [{
"TCP": "foo",
"Interval": 10,
"Status": "unsupported-status"
}]
}
}
}
`,
wantNS: nil,
wantErr: "Status for checks must 'passing', 'warning', 'critical'",
},
{
name: "another service registered with same ID as a sidecar should not be deregistered",
// Add another service with the same ID that a sidecar for web would have
preRegister: &structs.NodeService{
ID: "web-sidecar-proxy",
Service: "fake-sidecar",
Port: 9999,
},
// Register web with NO SIDECAR
json: `
{
"name": "web",
"port": 1111
}
`,
// Note here that although the registration here didn't register it, we
// should still see the NodeService we pre-registered here.
wantNS: &structs.NodeService{
ID: "web-sidecar-proxy",
Service: "fake-sidecar",
Port: 9999,
TaggedAddresses: map[string]structs.ServiceAddress{},
Weights: &structs.Weights{
Passing: 1,
Warning: 1,
},
EnterpriseMeta: *structs.DefaultEnterpriseMetaInDefaultPartition(),
},
// After we deregister the web service above, the fake sidecar with
// clashing ID SHOULD NOT have been removed since it wasn't part of the
// original registration.
wantSidecarIDLeftAfterDereg: true,
},
{
name: "updates to sidecar should work",
// Add a valid sidecar already registered
preRegister: &structs.NodeService{
ID: "web-sidecar-proxy",
Service: "web-sidecar-proxy",
LocallyRegisteredAsSidecar: true,
Port: 9999,
},
// Register web with Sidecar on different port
json: `
{
"name": "web",
"port": 1111,
"connect": {
"SidecarService": {
"Port": 6666
}
}
}
`,
// Note here that although the registration here didn't register it, we
// should still see the NodeService we pre-registered here.
wantNS: &structs.NodeService{
Kind: "connect-proxy",
ID: "web-sidecar-proxy",
Service: "web-sidecar-proxy",
LocallyRegisteredAsSidecar: true,
Port: 6666,
TaggedAddresses: map[string]structs.ServiceAddress{},
Weights: &structs.Weights{
Passing: 1,
Warning: 1,
},
Proxy: structs.ConnectProxyConfig{
DestinationServiceName: "web",
DestinationServiceID: "web",
LocalServiceAddress: "127.0.0.1",
LocalServicePort: 1111,
},
EnterpriseMeta: *structs.DefaultEnterpriseMetaInDefaultPartition(),
},
},
{
name: "update that removes sidecar should NOT deregister it",
// Add web with a valid sidecar already registered
preRegister: &structs.NodeService{
ID: "web",
Service: "web",
Port: 1111,
},
preRegister2: testDefaultSidecar("web", 1111),
// Register (update) web and remove sidecar (and port for sanity check)
json: `
{
"name": "web",
"port": 2222
}
`,
// Sidecar should still be there such that API can update registration
// without accidentally removing a sidecar. This is equivalent to embedded
// checks which are not removed by just not being included in an update.
// We will document that sidecar registrations via API must be explicitiy
// deregistered.
wantNS: testDefaultSidecar("web", 1111),
// Sanity check the rest of the update happened though.
assertStateFn: func(t *testing.T, state *local.State) {
svc := state.Service(structs.NewServiceID("web", nil))
require.NotNil(t, svc)
require.Equal(t, 2222, svc.Port)
},
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
// Constrain auto ports to 1 available to make it deterministic
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
hcl := `ports {
sidecar_min_port = 2222
sidecar_max_port = 2222
}
`
if tt.enableACL {
hcl = hcl + TestACLConfig()
}
a := NewTestAgent(t, hcl+" "+extraHCL)
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
if tt.preRegister != nil {
require.NoError(t, a.addServiceFromSource(tt.preRegister, nil, false, "", ConfigSourceLocal))
}
if tt.preRegister2 != nil {
require.NoError(t, a.addServiceFromSource(tt.preRegister2, nil, false, "", ConfigSourceLocal))
}
// Create an ACL token with require policy
var token string
if tt.enableACL && tt.tokenRules != "" {
token = testCreateToken(t, a, tt.tokenRules)
}
br := bytes.NewBufferString(tt.json)
req, _ := http.NewRequest("PUT", "/v1/agent/service/register?token="+token, br)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if tt.wantErr != "" {
require.Contains(t, strings.ToLower(resp.Body.String()), strings.ToLower(tt.wantErr))
return
}
require.Equal(t, 200, resp.Code, "request failed with body: %s",
resp.Body.String())
// Sanity the target service registration
svcs := a.State.AllServices()
// Parse the expected definition into a ServiceDefinition
var sd structs.ServiceDefinition
err := json.Unmarshal([]byte(tt.json), &sd)
require.NoError(t, err)
require.NotEmpty(t, sd.Name)
svcID := sd.ID
if svcID == "" {
svcID = sd.Name
}
sid := structs.NewServiceID(svcID, nil)
svc, ok := svcs[sid]
require.True(t, ok, "has service "+sid.String())
assert.Equal(t, sd.Name, svc.Service)
assert.Equal(t, sd.Port, svc.Port)
// Ensure that the actual registered service _doesn't_ still have it's
// sidecar info since it's duplicate and we don't want that synced up to
// the catalog or included in responses particularly - it's just
// registration syntax sugar.
assert.Nil(t, svc.Connect.SidecarService)
if tt.wantNS == nil {
// Sanity check that there was no service registered, we rely on there
// being no services at start of test so we can just use the count.
assert.Len(t, svcs, 1, "should be no sidecar registered")
return
}
// Ensure sidecar
svc, ok = svcs[structs.NewServiceID(tt.wantNS.ID, nil)]
require.True(t, ok, "no sidecar registered at "+tt.wantNS.ID)
assert.Equal(t, tt.wantNS, svc)
if tt.assertStateFn != nil {
tt.assertStateFn(t, a.State)
}
// Now verify deregistration also removes sidecar (if there was one and it
// was added via sidecar not just coincidental ID clash)
{
req := httptest.NewRequest("PUT",
"/v1/agent/service/deregister/"+svcID+"?token="+token, nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
svcs := a.State.AllServices()
_, ok = svcs[structs.NewServiceID(tt.wantNS.ID, nil)]
if tt.wantSidecarIDLeftAfterDereg {
require.True(t, ok, "removed non-sidecar service at "+tt.wantNS.ID)
} else {
require.False(t, ok, "sidecar not deregistered with service "+svcID)
}
}
})
}
}
2018-03-12 20:05:06 +00:00
// This tests that connect proxy validation is done for local agent
// registration. This doesn't need to test validation exhaustively since
// that is done via a table test in the structs package.
func TestAgent_RegisterService_UnmanagedConnectProxyInvalid(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Run("normal", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_UnmanagedConnectProxyInvalid(t, "enable_central_service_config = false")
})
t.Run("service manager", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_UnmanagedConnectProxyInvalid(t, "enable_central_service_config = true")
})
}
func testAgent_RegisterService_UnmanagedConnectProxyInvalid(t *testing.T, extraHCL string) {
t.Helper()
a := NewTestAgent(t, extraHCL)
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
args := &structs.ServiceDefinition{
Add Proxy Upstreams to Service Definition (#4639) * Refactor Service Definition ProxyDestination. This includes: - Refactoring all internal structs used - Updated tests for both deprecated and new input for: - Agent Services endpoint response - Agent Service endpoint response - Agent Register endpoint - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Register - Unmanaged deprecated field - Unmanaged new fields - Managed deprecated upstreams - Managed new - Catalog Services endpoint response - Catalog Node endpoint response - Catalog Service endpoint response - Updated API tests for all of the above too (both deprecated and new forms of register) TODO: - config package changes for on-disk service definitions - proxy config endpoint - built-in proxy support for new fields * Agent proxy config endpoint updated with upstreams * Config file changes for upstreams. * Add upstream opaque config and update all tests to ensure it works everywhere. * Built in proxy working with new Upstreams config * Command fixes and deprecations * Fix key translation, upstream type defaults and a spate of other subtele bugs found with ned to end test scripts... TODO: tests still failing on one case that needs a fix. I think it's key translation for upstreams nested in Managed proxy struct. * Fix translated keys in API registration. ≈ * Fixes from docs - omit some empty undocumented fields in API - Bring back ServiceProxyDestination in Catalog responses to not break backwards compat - this was removed assuming it was only used internally. * Documentation updates for Upstreams in service definition * Fixes for tests broken by many refactors. * Enable travis on f-connect branch in this branch too. * Add consistent Deprecation comments to ProxyDestination uses * Update version number on deprecation notices, and correct upstream datacenter field with explanation in docs
2018-09-12 16:07:47 +00:00
Kind: structs.ServiceKindConnectProxy,
Name: "connect-proxy",
Proxy: &structs.ConnectProxyConfig{
DestinationServiceName: "db",
},
Check: structs.CheckType{
TTL: 15 * time.Second,
},
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register?token=abc123", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
assert.Equal(t, http.StatusBadRequest, resp.Code)
assert.Contains(t, resp.Body.String(), "Port")
// Ensure the service doesn't exist
assert.Nil(t, a.State.Service(structs.NewServiceID("connect-proxy", nil)))
}
// Tests agent registration of a service that is connect native.
func TestAgent_RegisterService_ConnectNative(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Run("normal", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_ConnectNative(t, "enable_central_service_config = false")
})
t.Run("service manager", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_ConnectNative(t, "enable_central_service_config = true")
})
}
func testAgent_RegisterService_ConnectNative(t *testing.T, extraHCL string) {
t.Helper()
a := NewTestAgent(t, extraHCL)
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
// Register a proxy. Note that the destination doesn't exist here on
// this agent or in the catalog at all. This is intended and part
// of the design.
args := &structs.ServiceDefinition{
Name: "web",
Port: 8000,
Check: structs.CheckType{
TTL: 15 * time.Second,
},
Connect: &structs.ServiceConnect{
Native: true,
},
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
assert.Equal(t, http.StatusOK, resp.Code)
// Ensure the service
svc := a.State.Service(structs.NewServiceID("web", nil))
require.NotNil(t, svc)
assert.True(t, svc.Connect.Native)
}
func TestAgent_RegisterService_ScriptCheck_ExecDisable(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Run("normal", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_ScriptCheck_ExecDisable(t, "enable_central_service_config = false")
})
t.Run("service manager", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_ScriptCheck_ExecDisable(t, "enable_central_service_config = true")
})
}
func testAgent_RegisterService_ScriptCheck_ExecDisable(t *testing.T, extraHCL string) {
t.Helper()
a := NewTestAgent(t, extraHCL)
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
args := &structs.ServiceDefinition{
Name: "test",
Meta: map[string]string{"hello": "world"},
Tags: []string{"primary"},
Port: 8000,
Check: structs.CheckType{
Name: "test-check",
Interval: time.Second,
ScriptArgs: []string{"true"},
},
Weights: &structs.Weights{
Passing: 100,
Warning: 3,
},
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register?token=abc123", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if http.StatusInternalServerError != resp.Code {
t.Fatalf("expected 500 but got %v", resp.Code)
}
if !strings.Contains(resp.Body.String(), "Scripts are disabled on this agent") {
t.Fatalf("expected script disabled error, got: %s", resp.Body.String())
}
checkID := types.CheckID("test-check")
require.Nil(t, a.State.Check(structs.NewCheckID(checkID, nil)), "check registered with exec disabled")
}
func TestAgent_RegisterService_ScriptCheck_ExecRemoteDisable(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Run("normal", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_ScriptCheck_ExecRemoteDisable(t, "enable_central_service_config = false")
})
t.Run("service manager", func(t *testing.T) {
t.Parallel()
testAgent_RegisterService_ScriptCheck_ExecRemoteDisable(t, "enable_central_service_config = true")
})
}
func testAgent_RegisterService_ScriptCheck_ExecRemoteDisable(t *testing.T, extraHCL string) {
t.Helper()
a := NewTestAgent(t, `
enable_local_script_checks = true
`+extraHCL)
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
args := &structs.ServiceDefinition{
Name: "test",
Meta: map[string]string{"hello": "world"},
Tags: []string{"primary"},
Port: 8000,
Check: structs.CheckType{
Name: "test-check",
Interval: time.Second,
ScriptArgs: []string{"true"},
},
Weights: &structs.Weights{
Passing: 100,
Warning: 3,
},
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register?token=abc123", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if http.StatusInternalServerError != resp.Code {
t.Fatalf("expected 500 but got %v", resp.Code)
}
if !strings.Contains(resp.Body.String(), "Scripts are disabled on this agent") {
t.Fatalf("expected script disabled error, got: %s", resp.Body.String())
}
checkID := types.CheckID("test-check")
require.Nil(t, a.State.Check(structs.NewCheckID(checkID, nil)), "check registered with exec disabled")
}
func TestAgent_DeregisterService(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
serviceReq := AddServiceRequest{
Service: &structs.NodeService{
ID: "test",
Service: "test",
},
chkTypes: nil,
persist: false,
token: "",
Source: ConfigSourceLocal,
}
require.NoError(t, a.AddService(serviceReq))
req, _ := http.NewRequest("PUT", "/v1/agent/service/deregister/test", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if http.StatusOK != resp.Code {
t.Fatalf("expected 200 but got %v", resp.Code)
}
// Ensure we have a check mapping
assert.Nil(t, a.State.Service(structs.NewServiceID("test", nil)), "have test service")
assert.Nil(t, a.State.Check(structs.NewCheckID("test", nil)), "have test check")
}
func TestAgent_DeregisterService_ACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
serviceReq := AddServiceRequest{
Service: &structs.NodeService{
ID: "test",
Service: "test",
},
chkTypes: nil,
persist: false,
token: "",
Source: ConfigSourceLocal,
}
require.NoError(t, a.AddService(serviceReq))
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/service/deregister/test", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("root token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/service/deregister/test?token=root", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
}
func TestAgent_ServiceMaintenance_BadRequest(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
t.Run("not enabled", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/service/maintenance/test", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if resp.Code != 400 {
t.Fatalf("expected 400, got %d", resp.Code)
}
})
t.Run("no service id", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/service/maintenance/?enable=true", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if resp.Code != 400 {
t.Fatalf("expected 400, got %d", resp.Code)
}
})
t.Run("bad service id", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/service/maintenance/_nope_?enable=true", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, 404, resp.Code)
2022-01-05 20:18:08 +00:00
sid := structs.NewServiceID("_nope_", nil)
require.Contains(t, resp.Body.String(), fmt.Sprintf(`Unknown service ID %q`, sid))
})
}
func TestAgent_ServiceMaintenance_Enable(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
// Register the service
serviceReq := AddServiceRequest{
Service: &structs.NodeService{
ID: "test",
Service: "test",
},
chkTypes: nil,
persist: false,
token: "",
Source: ConfigSourceLocal,
}
require.NoError(t, a.AddService(serviceReq))
// Force the service into maintenance mode
req, _ := http.NewRequest("PUT", "/v1/agent/service/maintenance/test?enable=true&reason=broken&token=mytoken", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if resp.Code != 200 {
t.Fatalf("expected 200, got %d", resp.Code)
}
// Ensure the maintenance check was registered
checkID := serviceMaintCheckID(structs.NewServiceID("test", nil))
check := a.State.Check(checkID)
if check == nil {
t.Fatalf("should have registered maintenance check")
}
// Ensure the token was added
2017-08-28 12:17:13 +00:00
if token := a.State.CheckToken(checkID); token != "mytoken" {
t.Fatalf("expected 'mytoken', got '%s'", token)
}
// Ensure the reason was set in notes
if check.Notes != "broken" {
t.Fatalf("bad: %#v", check)
}
}
func TestAgent_ServiceMaintenance_Disable(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
// Register the service
serviceReq := AddServiceRequest{
Service: &structs.NodeService{
ID: "test",
Service: "test",
},
chkTypes: nil,
persist: false,
token: "",
Source: ConfigSourceLocal,
}
require.NoError(t, a.AddService(serviceReq))
// Force the service into maintenance mode
if err := a.EnableServiceMaintenance(structs.NewServiceID("test", nil), "", ""); err != nil {
t.Fatalf("err: %s", err)
}
// Leave maintenance mode
req, _ := http.NewRequest("PUT", "/v1/agent/service/maintenance/test?enable=false", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if resp.Code != 200 {
t.Fatalf("expected 200, got %d", resp.Code)
}
// Ensure the maintenance check was removed
checkID := serviceMaintCheckID(structs.NewServiceID("test", nil))
if existing := a.State.Check(checkID); existing != nil {
t.Fatalf("should have removed maintenance check")
}
}
2015-01-15 19:26:14 +00:00
func TestAgent_ServiceMaintenance_ACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
// Register the service.
serviceReq := AddServiceRequest{
Service: &structs.NodeService{
ID: "test",
Service: "test",
},
chkTypes: nil,
persist: false,
token: "",
Source: ConfigSourceLocal,
}
require.NoError(t, a.AddService(serviceReq))
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/service/maintenance/test?enable=true&reason=broken", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("root token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/service/maintenance/test?enable=true&reason=broken&token=root", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
}
func TestAgent_NodeMaintenance_BadRequest(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
2015-01-15 19:26:14 +00:00
// Fails when no enable flag provided
req, _ := http.NewRequest("PUT", "/v1/agent/maintenance", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
2015-01-15 19:26:14 +00:00
if resp.Code != 400 {
t.Fatalf("expected 400, got %d", resp.Code)
}
}
func TestAgent_NodeMaintenance_Enable(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
2015-01-15 19:26:14 +00:00
// Force the node into maintenance mode
req, _ := http.NewRequest("PUT", "/v1/agent/maintenance?enable=true&reason=broken&token=mytoken", nil)
2015-01-15 19:26:14 +00:00
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
2015-01-15 19:26:14 +00:00
if resp.Code != 200 {
t.Fatalf("expected 200, got %d", resp.Code)
}
// Ensure the maintenance check was registered
check := a.State.Check(structs.NodeMaintCheckID)
if check == nil {
2015-01-15 19:26:14 +00:00
t.Fatalf("should have registered maintenance check")
}
// Check that the token was used
if token := a.State.CheckToken(structs.NodeMaintCheckID); token != "mytoken" {
t.Fatalf("expected 'mytoken', got '%s'", token)
}
// Ensure the reason was set in notes
if check.Notes != "broken" {
t.Fatalf("bad: %#v", check)
}
2015-01-15 19:26:14 +00:00
}
func TestAgent_NodeMaintenance_Disable(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
2015-01-15 19:26:14 +00:00
// Force the node into maintenance mode
a.EnableNodeMaintenance("", "")
2015-01-15 19:26:14 +00:00
// Leave maintenance mode
req, _ := http.NewRequest("PUT", "/v1/agent/maintenance?enable=false", nil)
2015-01-15 19:26:14 +00:00
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
2015-01-15 19:26:14 +00:00
if resp.Code != 200 {
t.Fatalf("expected 200, got %d", resp.Code)
}
// Ensure the maintenance check was removed
if existing := a.State.Check(structs.NodeMaintCheckID); existing != nil {
2015-01-15 19:26:14 +00:00
t.Fatalf("should have removed maintenance check")
}
}
func TestAgent_NodeMaintenance_ACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
t.Run("no token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/maintenance?enable=true&reason=broken", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
})
t.Run("root token", func(t *testing.T) {
req, _ := http.NewRequest("PUT", "/v1/agent/maintenance?enable=true&reason=broken&token=root", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
})
}
func TestAgent_RegisterCheck_Service(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
args := &structs.ServiceDefinition{
Name: "memcache",
Port: 8000,
Check: structs.CheckType{
TTL: 15 * time.Second,
},
}
2017-05-09 16:58:12 +00:00
// First register the service
req, _ := http.NewRequest("PUT", "/v1/agent/service/register", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if resp.Code != 200 {
t.Fatalf("expected 200, got %d", resp.Code)
}
// Now register an additional check
checkArgs := &structs.CheckDefinition{
Name: "memcache_check2",
ServiceID: "memcache",
TTL: 15 * time.Second,
}
req, _ = http.NewRequest("PUT", "/v1/agent/check/register", jsonReader(checkArgs))
resp = httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if resp.Code != 200 {
t.Fatalf("expected 200, got %d", resp.Code)
}
// Ensure we have a check mapping
result := a.State.Checks(nil)
if _, ok := result[structs.NewCheckID("service:memcache", nil)]; !ok {
t.Fatalf("missing memcached check")
}
if _, ok := result[structs.NewCheckID("memcache_check2", nil)]; !ok {
t.Fatalf("missing memcache_check2 check")
}
// Make sure the new check is associated with the service
if result[structs.NewCheckID("memcache_check2", nil)].ServiceID != "memcache" {
t.Fatalf("bad: %#v", result[structs.NewCheckID("memcached_check2", nil)])
}
2019-10-17 18:33:11 +00:00
// Make sure the new check has the right type
if result[structs.NewCheckID("memcache_check2", nil)].Type != "ttl" {
t.Fatalf("expected TTL type, got %s", result[structs.NewCheckID("memcache_check2", nil)].Type)
2019-10-17 18:33:11 +00:00
}
}
2016-11-16 21:45:26 +00:00
func TestAgent_Monitor(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
2016-11-16 21:45:26 +00:00
t.Run("unknown log level", func(t *testing.T) {
// Try passing an invalid log level
req, _ := http.NewRequest("GET", "/v1/agent/monitor?loglevel=invalid", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if http.StatusBadRequest != resp.Code {
t.Fatalf("expected 400 but got %v", resp.Code)
}
substring := "Unknown log level"
if !strings.Contains(resp.Body.String(), substring) {
t.Fatalf("got: %s, wanted message containing: %s", resp.Body.String(), substring)
}
})
t.Run("stream unstructured logs", func(t *testing.T) {
// Try to stream logs until we see the expected log line
retry.Run(t, func(r *retry.R) {
req, _ := http.NewRequest("GET", "/v1/agent/monitor?loglevel=debug", nil)
cancelCtx, cancelFunc := context.WithCancel(context.Background())
req = req.WithContext(cancelCtx)
resp := httptest.NewRecorder()
codeCh := make(chan int)
go func() {
a.srv.h.ServeHTTP(resp, req)
codeCh <- resp.Code
}()
args := &structs.ServiceDefinition{
Name: "monitor",
Port: 8000,
Check: structs.CheckType{
TTL: 15 * time.Second,
},
2016-11-16 21:45:26 +00:00
}
registerReq, _ := http.NewRequest("PUT", "/v1/agent/service/register", jsonReader(args))
res := httptest.NewRecorder()
a.srv.h.ServeHTTP(res, registerReq)
if http.StatusOK != res.Code {
t.Fatalf("expected 200 but got %v", res.Code)
}
// Wait until we have received some type of logging output
require.Eventually(t, func() bool {
return len(resp.Body.Bytes()) > 0
}, 3*time.Second, 100*time.Millisecond)
cancelFunc()
code := <-codeCh
require.Equal(t, http.StatusOK, code)
got := resp.Body.String()
// Only check a substring that we are highly confident in finding
want := "Synced service: service="
if !strings.Contains(got, want) {
r.Fatalf("got %q and did not find %q", got, want)
}
})
})
2016-11-16 21:45:26 +00:00
t.Run("stream compressed unstructured logs", func(t *testing.T) {
// The only purpose of this test is to see something being
// logged. Because /v1/agent/monitor is streaming the response
// it needs special handling with the compression.
retry.Run(t, func(r *retry.R) {
req, _ := http.NewRequest("GET", "/v1/agent/monitor?loglevel=debug", nil)
// Usually this would be automatically set by transport content
// negotiation, but since this call doesn't go through a real
// transport, the header has to be set manually
req.Header["Accept-Encoding"] = []string{"gzip"}
cancelCtx, cancelFunc := context.WithCancel(context.Background())
req = req.WithContext(cancelCtx)
resp := httptest.NewRecorder()
handler := a.srv.handler(true)
go handler.ServeHTTP(resp, req)
args := &structs.ServiceDefinition{
Name: "monitor",
Port: 8000,
Check: structs.CheckType{
TTL: 15 * time.Second,
},
}
registerReq, _ := http.NewRequest("PUT", "/v1/agent/service/register", jsonReader(args))
res := httptest.NewRecorder()
a.srv.h.ServeHTTP(res, registerReq)
if http.StatusOK != res.Code {
t.Fatalf("expected 200 but got %v", res.Code)
}
// Wait until we have received some type of logging output
require.Eventually(t, func() bool {
return len(resp.Body.Bytes()) > 0
}, 3*time.Second, 100*time.Millisecond)
cancelFunc()
})
})
t.Run("stream JSON logs", func(t *testing.T) {
// Try to stream logs until we see the expected log line
retry.Run(t, func(r *retry.R) {
req, _ := http.NewRequest("GET", "/v1/agent/monitor?loglevel=debug&logjson", nil)
cancelCtx, cancelFunc := context.WithCancel(context.Background())
req = req.WithContext(cancelCtx)
2016-11-16 21:45:26 +00:00
resp := httptest.NewRecorder()
codeCh := make(chan int)
go func() {
a.srv.h.ServeHTTP(resp, req)
codeCh <- resp.Code
}()
2016-11-16 21:45:26 +00:00
args := &structs.ServiceDefinition{
Name: "monitor",
Port: 8000,
Check: structs.CheckType{
TTL: 15 * time.Second,
},
}
registerReq, _ := http.NewRequest("PUT", "/v1/agent/service/register", jsonReader(args))
res := httptest.NewRecorder()
a.srv.h.ServeHTTP(res, registerReq)
if http.StatusOK != res.Code {
t.Fatalf("expected 200 but got %v", res.Code)
}
// Wait until we have received some type of logging output
require.Eventually(t, func() bool {
return len(resp.Body.Bytes()) > 0
}, 3*time.Second, 100*time.Millisecond)
cancelFunc()
code := <-codeCh
require.Equal(t, http.StatusOK, code)
// Each line is output as a separate JSON object, we grab the first and
// make sure it can be unmarshalled.
firstLine := bytes.Split(resp.Body.Bytes(), []byte("\n"))[0]
var output map[string]interface{}
if err := json.Unmarshal(firstLine, &output); err != nil {
t.Fatalf("err: %v", err)
}
})
})
// hopefully catch any potential regression in serf/memberlist logging setup.
t.Run("serf shutdown logging", func(t *testing.T) {
req, _ := http.NewRequest("GET", "/v1/agent/monitor?loglevel=debug", nil)
cancelCtx, cancelFunc := context.WithCancel(context.Background())
req = req.WithContext(cancelCtx)
resp := httptest.NewRecorder()
codeCh := make(chan int)
chStarted := make(chan struct{})
go func() {
close(chStarted)
a.srv.h.ServeHTTP(resp, req)
codeCh <- resp.Code
}()
<-chStarted
require.NoError(t, a.Shutdown())
// Wait until we have received some type of logging output
require.Eventually(t, func() bool {
return len(resp.Body.Bytes()) > 0
}, 3*time.Second, 100*time.Millisecond)
cancelFunc()
code := <-codeCh
require.Equal(t, http.StatusOK, code)
got := resp.Body.String()
want := "serf: Shutdown without a Leave"
if !strings.Contains(got, want) {
t.Fatalf("got %q and did not find %q", got, want)
}
})
2016-11-16 21:45:26 +00:00
}
func TestAgent_Monitor_ACLDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
// Try without a token.
req, _ := http.NewRequest("GET", "/v1/agent/monitor", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if http.StatusForbidden != resp.Code {
t.Fatalf("expected 403 but got %v", resp.Code)
}
// This proves we call the ACL function, and we've got the other monitor
// test to prove monitor works, which should be sufficient. The monitor
// logic is a little complex to set up so isn't worth repeating again
// here.
}
func TestAgent_TokenTriggersFullSync(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
body := func(token string) io.Reader {
return jsonReader(&api.AgentToken{Token: token})
}
createNodePolicy := func(t *testing.T, a *TestAgent, policyName string) *structs.ACLPolicy {
policy := &structs.ACLPolicy{
Name: policyName,
Rules: `node_prefix "" { policy = "write" }`,
}
req, err := http.NewRequest("PUT", "/v1/acl/policy?token=root", jsonBody(policy))
require.NoError(t, err)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
dec := json.NewDecoder(resp.Body)
policy = &structs.ACLPolicy{}
require.NoError(t, dec.Decode(policy))
return policy
}
createNodeToken := func(t *testing.T, a *TestAgent, policyName string) *structs.ACLToken {
createNodePolicy(t, a, policyName)
token := &structs.ACLToken{
Description: "test",
Policies: []structs.ACLTokenPolicyLink{
{Name: policyName},
},
}
req, err := http.NewRequest("PUT", "/v1/acl/token?token=root", jsonBody(token))
require.NoError(t, err)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
dec := json.NewDecoder(resp.Body)
token = &structs.ACLToken{}
require.NoError(t, dec.Decode(token))
return token
}
cases := []struct {
path string
tokenGetFn func(*token.Store) string
}{
{
path: "acl_agent_token",
tokenGetFn: (*token.Store).AgentToken,
},
{
path: "agent",
tokenGetFn: (*token.Store).AgentToken,
},
{
path: "acl_token",
tokenGetFn: (*token.Store).UserToken,
},
{
path: "default",
tokenGetFn: (*token.Store).UserToken,
},
}
for _, tt := range cases {
tt := tt
t.Run(tt.path, func(t *testing.T) {
url := fmt.Sprintf("/v1/agent/token/%s?token=root", tt.path)
a := NewTestAgent(t, `
primary_datacenter = "dc1"
acl {
enabled = true
default_policy = "deny"
tokens {
initial_management = "root"
default = ""
agent = ""
agent_recovery = ""
replication = ""
}
}
`)
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
// create node policy and token
token := createNodeToken(t, a, "test")
req, err := http.NewRequest("PUT", url, body(token.SecretID))
require.NoError(t, err)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
require.Equal(t, token.SecretID, tt.tokenGetFn(a.tokens))
testrpc.WaitForTestAgent(t, a.RPC, "dc1",
testrpc.WithToken("root"),
testrpc.WaitForAntiEntropySync())
})
}
}
func TestAgent_Token(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
// The behavior of this handler when ACLs are disabled is vetted over
// in TestACL_Disabled_Response since there's already good infra set
// up over there to test this, and it calls the common function.
a := NewTestAgent(t, `
primary_datacenter = "dc1"
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
acl {
enabled = true
default_policy = "deny"
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
tokens {
initial_management = "root"
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
default = ""
agent = ""
agent_recovery = ""
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
replication = ""
}
}
New config parser, HCL support, multiple bind addrs (#3480) * new config parser for agent This patch implements a new config parser for the consul agent which makes the following changes to the previous implementation: * add HCL support * all configuration fragments in tests and for default config are expressed as HCL fragments * HCL fragments can be provided on the command line so that they can eventually replace the command line flags. * HCL/JSON fragments are parsed into a temporary Config structure which can be merged using reflection (all values are pointers). The existing merge logic of overwrite for values and append for slices has been preserved. * A single builder process generates a typed runtime configuration for the agent. The new implementation is more strict and fails in the builder process if no valid runtime configuration can be generated. Therefore, additional validations in other parts of the code should be removed. The builder also pre-computes all required network addresses so that no address/port magic should be required where the configuration is used and should therefore be removed. * Upgrade github.com/hashicorp/hcl to support int64 * improve error messages * fix directory permission test * Fix rtt test * Fix ForceLeave test * Skip performance test for now until we know what to do * Update github.com/hashicorp/memberlist to update log prefix * Make memberlist use the default logger * improve config error handling * do not fail on non-existing data-dir * experiment with non-uniform timeouts to get a handle on stalled leader elections * Run tests for packages separately to eliminate the spurious port conflicts * refactor private address detection and unify approach for ipv4 and ipv6. Fixes #2825 * do not allow unix sockets for DNS * improve bind and advertise addr error handling * go through builder using test coverage * minimal update to the docs * more coverage tests fixed * more tests * fix makefile * cleanup * fix port conflicts with external port server 'porter' * stop test server on error * do not run api test that change global ENV concurrently with the other tests * Run remaining api tests concurrently * no need for retry with the port number service * monkey patch race condition in go-sockaddr until we understand why that fails * monkey patch hcl decoder race condidtion until we understand why that fails * monkey patch spurious errors in strings.EqualFold from here * add test for hcl decoder race condition. Run with go test -parallel 128 * Increase timeout again * cleanup * don't log port allocations by default * use base command arg parsing to format help output properly * handle -dc deprecation case in Build * switch autopilot.max_trailing_logs to int * remove duplicate test case * remove unused methods * remove comments about flag/config value inconsistencies * switch got and want around since the error message was misleading. * Removes a stray debug log. * Removes a stray newline in imports. * Fixes TestACL_Version8. * Runs go fmt. * Adds a default case for unknown address types. * Reoders and reformats some imports. * Adds some comments and fixes typos. * Reorders imports. * add unix socket support for dns later * drop all deprecated flags and arguments * fix wrong field name * remove stray node-id file * drop unnecessary patch section in test * drop duplicate test * add test for LeaveOnTerm and SkipLeaveOnInt in client mode * drop "bla" and add clarifying comment for the test * split up tests to support enterprise/non-enterprise tests * drop raft multiplier and derive values during build phase * sanitize runtime config reflectively and add test * detect invalid config fields * fix tests with invalid config fields * use different values for wan sanitiziation test * drop recursor in favor of recursors * allow dns_config.udp_answer_limit to be zero * make sure tests run on machines with multiple ips * Fix failing tests in a few more places by providing a bind address in the test * Gets rid of skipped TestAgent_CheckPerformanceSettings and adds case for builder. * Add porter to server_test.go to make tests there less flaky * go fmt
2017-09-25 18:40:42 +00:00
`)
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
type tokens struct {
user string
userSource tokenStore.TokenSource
agent string
agentSource tokenStore.TokenSource
agentRecovery string
agentRecoverySource tokenStore.TokenSource
repl string
replSource tokenStore.TokenSource
}
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
resetTokens := func(init tokens) {
a.tokens.UpdateUserToken(init.user, init.userSource)
a.tokens.UpdateAgentToken(init.agent, init.agentSource)
a.tokens.UpdateAgentRecoveryToken(init.agentRecovery, init.agentRecoverySource)
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
a.tokens.UpdateReplicationToken(init.repl, init.replSource)
}
body := func(token string) io.Reader {
return jsonReader(&api.AgentToken{Token: token})
}
badJSON := func() io.Reader {
return jsonReader(false)
}
tests := []struct {
name string
method, url string
body io.Reader
code int
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
init tokens
raw tokens
effective tokens
expectedErr string
}{
{
name: "bad token name",
method: "PUT",
url: "nope?token=root",
body: body("X"),
code: http.StatusNotFound,
expectedErr: `Token "nope" is unknown`,
},
{
name: "bad JSON",
method: "PUT",
url: "acl_token?token=root",
body: badJSON(),
code: http.StatusBadRequest,
expectedErr: `Bad request: Request decode failed: json: cannot unmarshal bool into Go value of type api.AgentToken`,
},
{
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
name: "set user legacy",
method: "PUT",
url: "acl_token?token=root",
body: body("U"),
code: http.StatusOK,
raw: tokens{user: "U", userSource: tokenStore.TokenSourceAPI},
effective: tokens{user: "U", agent: "U"},
},
{
name: "set default",
method: "PUT",
url: "default?token=root",
body: body("U"),
code: http.StatusOK,
raw: tokens{user: "U", userSource: tokenStore.TokenSourceAPI},
effective: tokens{user: "U", agent: "U"},
},
{
name: "set agent legacy",
method: "PUT",
url: "acl_agent_token?token=root",
body: body("A"),
code: http.StatusOK,
init: tokens{user: "U", agent: "U"},
raw: tokens{user: "U", agent: "A", agentSource: tokenStore.TokenSourceAPI},
effective: tokens{user: "U", agent: "A"},
},
{
name: "set agent",
method: "PUT",
url: "agent?token=root",
body: body("A"),
code: http.StatusOK,
init: tokens{user: "U", agent: "U"},
raw: tokens{user: "U", agent: "A", agentSource: tokenStore.TokenSourceAPI},
effective: tokens{user: "U", agent: "A"},
},
{
name: "set master legacy",
method: "PUT",
url: "acl_agent_master_token?token=root",
body: body("M"),
code: http.StatusOK,
raw: tokens{agentRecovery: "M", agentRecoverySource: tokenStore.TokenSourceAPI},
effective: tokens{agentRecovery: "M"},
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
},
{
name: "set master",
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
method: "PUT",
url: "agent_master?token=root",
body: body("M"),
code: http.StatusOK,
raw: tokens{agentRecovery: "M", agentRecoverySource: tokenStore.TokenSourceAPI},
effective: tokens{agentRecovery: "M"},
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
},
{
name: "set recovery",
method: "PUT",
url: "agent_recovery?token=root",
body: body("R"),
code: http.StatusOK,
raw: tokens{agentRecovery: "R", agentRecoverySource: tokenStore.TokenSourceAPI},
effective: tokens{agentRecovery: "R", agentRecoverySource: tokenStore.TokenSourceAPI},
},
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
{
name: "set repl legacy",
method: "PUT",
url: "acl_replication_token?token=root",
body: body("R"),
code: http.StatusOK,
raw: tokens{repl: "R", replSource: tokenStore.TokenSourceAPI},
effective: tokens{repl: "R"},
},
{
name: "set repl",
method: "PUT",
url: "replication?token=root",
body: body("R"),
code: http.StatusOK,
raw: tokens{repl: "R", replSource: tokenStore.TokenSourceAPI},
effective: tokens{repl: "R"},
},
{
name: "clear user legacy",
method: "PUT",
url: "acl_token?token=root",
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
body: body(""),
code: http.StatusOK,
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
init: tokens{user: "U"},
raw: tokens{userSource: tokenStore.TokenSourceAPI},
},
{
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
name: "clear default",
method: "PUT",
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
url: "default?token=root",
body: body(""),
code: http.StatusOK,
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
init: tokens{user: "U"},
raw: tokens{userSource: tokenStore.TokenSourceAPI},
},
{
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
name: "clear agent legacy",
method: "PUT",
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
url: "acl_agent_token?token=root",
body: body(""),
code: http.StatusOK,
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
init: tokens{agent: "A"},
raw: tokens{agentSource: tokenStore.TokenSourceAPI},
},
{
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
name: "clear agent",
method: "PUT",
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
url: "agent?token=root",
body: body(""),
code: http.StatusOK,
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
init: tokens{agent: "A"},
raw: tokens{agentSource: tokenStore.TokenSourceAPI},
},
{
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
name: "clear master legacy",
method: "PUT",
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
url: "acl_agent_master_token?token=root",
body: body(""),
code: http.StatusOK,
init: tokens{agentRecovery: "M"},
raw: tokens{agentRecoverySource: tokenStore.TokenSourceAPI},
},
{
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
name: "clear master",
method: "PUT",
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
url: "agent_master?token=root",
body: body(""),
code: http.StatusOK,
init: tokens{agentRecovery: "M"},
raw: tokens{agentRecoverySource: tokenStore.TokenSourceAPI},
},
{
name: "clear recovery",
method: "PUT",
url: "agent_recovery?token=root",
body: body(""),
code: http.StatusOK,
init: tokens{agentRecovery: "R"},
raw: tokens{agentRecoverySource: tokenStore.TokenSourceAPI},
},
{
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
name: "clear repl legacy",
method: "PUT",
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
url: "acl_replication_token?token=root",
body: body(""),
code: http.StatusOK,
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
init: tokens{repl: "R"},
raw: tokens{replSource: tokenStore.TokenSourceAPI},
},
{
name: "clear repl",
method: "PUT",
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
url: "replication?token=root",
body: body(""),
code: http.StatusOK,
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
init: tokens{repl: "R"},
raw: tokens{replSource: tokenStore.TokenSourceAPI},
},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
resetTokens(tt.init)
url := fmt.Sprintf("/v1/agent/token/%s", tt.url)
resp := httptest.NewRecorder()
req, _ := http.NewRequest(tt.method, url, tt.body)
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, tt.code, resp.Code)
if tt.expectedErr != "" {
require.Contains(t, resp.Body.String(), tt.expectedErr)
return
}
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
require.Equal(t, tt.effective.user, a.tokens.UserToken())
require.Equal(t, tt.effective.agent, a.tokens.AgentToken())
require.Equal(t, tt.effective.agentRecovery, a.tokens.AgentRecoveryToken())
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
require.Equal(t, tt.effective.repl, a.tokens.ReplicationToken())
tok, src := a.tokens.UserTokenAndSource()
require.Equal(t, tt.raw.user, tok)
require.Equal(t, tt.raw.userSource, src)
tok, src = a.tokens.AgentTokenAndSource()
require.Equal(t, tt.raw.agent, tok)
require.Equal(t, tt.raw.agentSource, src)
tok, src = a.tokens.AgentRecoveryTokenAndSource()
require.Equal(t, tt.raw.agentRecovery, tok)
require.Equal(t, tt.raw.agentRecoverySource, src)
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
tok, src = a.tokens.ReplicationTokenAndSource()
require.Equal(t, tt.raw.repl, tok)
require.Equal(t, tt.raw.replSource, src)
})
}
// This one returns an error that is interpreted by the HTTP wrapper, so
// doesn't fit into our table above.
t.Run("permission denied", func(t *testing.T) {
resetTokens(tokens{})
req, _ := http.NewRequest("PUT", "/v1/agent/token/acl_token", body("X"))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
ACL Token Persistence and Reloading (#5328) This PR adds two features which will be useful for operators when ACLs are in use. 1. Tokens set in configuration files are now reloadable. 2. If `acl.enable_token_persistence` is set to `true` in the configuration, tokens set via the `v1/agent/token` endpoint are now persisted to disk and loaded when the agent starts (or during configuration reload) Note that token persistence is opt-in so our users who do not want tokens on the local disk will see no change. Some other secondary changes: * Refactored a bunch of places where the replication token is retrieved from the token store. This token isn't just for replicating ACLs and now it is named accordingly. * Allowed better paths in the `v1/agent/token/` API. Instead of paths like: `v1/agent/token/acl_replication_token` the path can now be just `v1/agent/token/replication`. The old paths remain to be valid. * Added a couple new API functions to set tokens via the new paths. Deprecated the old ones and pointed to the new names. The names are also generally better and don't imply that what you are setting is for ACLs but rather are setting ACL tokens. There is a minor semantic difference there especially for the replication token as again, its no longer used only for ACL token/policy replication. The new functions will detect 404s and fallback to using the older token paths when talking to pre-1.4.3 agents. * Docs updated to reflect the API additions and to show using the new endpoints. * Updated the ACL CLI set-agent-tokens command to use the non-deprecated APIs.
2019-02-27 19:28:31 +00:00
require.Equal(t, "", a.tokens.UserToken())
})
}
2018-03-21 17:20:35 +00:00
func TestAgentConnectCARoots_empty(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
2018-03-21 17:20:35 +00:00
t.Parallel()
a := NewTestAgent(t, "connect { enabled = false }")
2018-03-21 17:20:35 +00:00
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
2018-03-21 17:20:35 +00:00
req, _ := http.NewRequest("GET", "/v1/agent/connect/ca/roots", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusInternalServerError, resp.Code)
require.Contains(t, resp.Body.String(), "Connect must be enabled")
2018-03-21 17:20:35 +00:00
}
func TestAgentConnectCARoots_list(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
2018-03-21 17:20:35 +00:00
t.Parallel()
a := NewTestAgent(t, "")
2018-03-21 17:20:35 +00:00
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
2018-03-21 17:20:35 +00:00
// Set some CAs. Note that NewTestAgent already bootstraps one CA so this just
// adds a second and makes it active.
ca2 := connect.TestCAConfigSet(t, a, nil)
2018-03-21 17:20:35 +00:00
// List
req, _ := http.NewRequest("GET", "/v1/agent/connect/ca/roots", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
value := &structs.IndexedCARoots{}
require.NoError(t, dec.Decode(value))
2018-03-21 17:20:35 +00:00
assert.Equal(t, value.ActiveRootID, ca2.ID)
2018-05-09 16:15:29 +00:00
// Would like to assert that it's the same as the TestAgent domain but the
// only way to access that state via this package is by RPC to the server
// implementation running in TestAgent which is more or less a tautology.
assert.NotEmpty(t, value.TrustDomain)
assert.Len(t, value.Roots, 2)
2018-03-21 17:20:35 +00:00
// We should never have the secret information
for _, r := range value.Roots {
assert.Equal(t, "", r.SigningCert)
assert.Equal(t, "", r.SigningKey)
}
assert.Equal(t, "MISS", resp.Header().Get("X-Cache"))
// Test caching
{
// List it again
resp2 := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp2, req)
dec := json.NewDecoder(resp2.Body)
value2 := &structs.IndexedCARoots{}
require.NoError(t, dec.Decode(value2))
assert.Equal(t, value, value2)
// Should cache hit this time and not make request
assert.Equal(t, "HIT", resp2.Header().Get("X-Cache"))
}
// Test that caching is updated in the background
{
// Set a new CA
ca := connect.TestCAConfigSet(t, a, nil)
retry.Run(t, func(r *retry.R) {
// List it again
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
value := &structs.IndexedCARoots{}
require.NoError(t, dec.Decode(value))
if ca.ID != value.ActiveRootID {
r.Fatalf("%s != %s", ca.ID, value.ActiveRootID)
}
// There are now 3 CAs because we didn't complete rotation on the original
// 2
if len(value.Roots) != 3 {
r.Fatalf("bad len: %d", len(value.Roots))
}
// Should be a cache hit! The data should've updated in the cache
// in the background so this should've been fetched directly from
// the cache.
if resp.Header().Get("X-Cache") != "HIT" {
r.Fatalf("should be a cache hit")
}
})
2018-03-21 17:20:35 +00:00
}
}
2018-03-21 17:55:39 +00:00
func TestAgentConnectCALeafCert_aclDefaultDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
testrpc.WaitForActiveCARoot(t, a.RPC, "dc1", nil)
// Register a service with a managed proxy
{
reg := &structs.ServiceDefinition{
ID: "test-id",
Name: "test",
Address: "127.0.0.1",
Port: 8000,
Check: structs.CheckType{
TTL: 15 * time.Second,
},
Connect: &structs.ServiceConnect{},
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register?token=root", jsonReader(reg))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, 200, resp.Code, "body: %s", resp.Body.String())
}
req, _ := http.NewRequest("GET", "/v1/agent/connect/ca/leaf/test", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
}
func TestAgentConnectCALeafCert_aclServiceWrite(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
testrpc.WaitForActiveCARoot(t, a.RPC, "dc1", nil)
// Register a service with a managed proxy
{
reg := &structs.ServiceDefinition{
ID: "test-id",
Name: "test",
Address: "127.0.0.1",
Port: 8000,
Check: structs.CheckType{
TTL: 15 * time.Second,
},
Connect: &structs.ServiceConnect{},
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register?token=root", jsonReader(reg))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, 200, resp.Code, "body: %s", resp.Body.String())
}
token := createACLTokenWithServicePolicy(t, a.srv, "write")
req, _ := http.NewRequest("GET", "/v1/agent/connect/ca/leaf/test?token="+token, nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
// Get the issued cert
dec := json.NewDecoder(resp.Body)
value := &structs.IssuedCert{}
require.NoError(t, dec.Decode(value))
require.NotNil(t, value)
}
func createACLTokenWithServicePolicy(t *testing.T, srv *HTTPHandlers, policy string) string {
policyReq := &structs.ACLPolicy{
Name: "service-test-write",
Rules: fmt.Sprintf(`service "test" { policy = "%v" }`, policy),
}
req, _ := http.NewRequest("PUT", "/v1/acl/policy?token=root", jsonReader(policyReq))
resp := httptest.NewRecorder()
_, err := srv.ACLPolicyCreate(resp, req)
require.NoError(t, err)
tokenReq := &structs.ACLToken{
Description: "token-for-test",
Policies: []structs.ACLTokenPolicyLink{{Name: "service-test-write"}},
}
req, _ = http.NewRequest("PUT", "/v1/acl/token?token=root", jsonReader(tokenReq))
resp = httptest.NewRecorder()
srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
svcToken := &structs.ACLToken{}
require.NoError(t, dec.Decode(svcToken))
return svcToken.SecretID
}
func TestAgentConnectCALeafCert_aclServiceReadDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
testrpc.WaitForActiveCARoot(t, a.RPC, "dc1", nil)
// Register a service with a managed proxy
{
reg := &structs.ServiceDefinition{
ID: "test-id",
Name: "test",
Address: "127.0.0.1",
Port: 8000,
Check: structs.CheckType{
TTL: 15 * time.Second,
},
Connect: &structs.ServiceConnect{},
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register?token=root", jsonReader(reg))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, 200, resp.Code, "body: %s", resp.Body.String())
}
token := createACLTokenWithServicePolicy(t, a.srv, "read")
req, _ := http.NewRequest("GET", "/v1/agent/connect/ca/leaf/test?token="+token, nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusForbidden, resp.Code)
}
2018-03-21 17:55:39 +00:00
func TestAgentConnectCALeafCert_good(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
2018-03-21 17:55:39 +00:00
t.Parallel()
a := StartTestAgent(t, TestAgent{Overrides: `
connect {
test_ca_leaf_root_change_spread = "1ns"
}
`})
2018-03-21 17:55:39 +00:00
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
testrpc.WaitForActiveCARoot(t, a.RPC, "dc1", nil)
2018-03-21 17:55:39 +00:00
// CA already setup by default by NewTestAgent but force a new one so we can
// verify it was signed easily.
ca1 := connect.TestCAConfigSet(t, a, nil)
2018-03-21 17:55:39 +00:00
{
// Register a local service
args := &structs.ServiceDefinition{
ID: "foo",
Name: "test",
Address: "127.0.0.1",
Port: 8000,
Check: structs.CheckType{
TTL: 15 * time.Second,
},
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if !assert.Equal(t, 200, resp.Code) {
2018-03-21 17:55:39 +00:00
t.Log("Body: ", resp.Body.String())
}
}
// List
req, _ := http.NewRequest("GET", "/v1/agent/connect/ca/leaf/test", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, "MISS", resp.Header().Get("X-Cache"))
2018-06-13 19:43:34 +00:00
// Get the issued cert
dec := json.NewDecoder(resp.Body)
issued := &structs.IssuedCert{}
require.NoError(t, dec.Decode(issued))
// Verify that the cert is signed by the CA
requireLeafValidUnderCA(t, issued, ca1)
// Verify blocking index
assert.True(t, issued.ModifyIndex > 0)
assert.Equal(t, fmt.Sprintf("%d", issued.ModifyIndex),
resp.Header().Get("X-Consul-Index"))
index := resp.Header().Get("X-Consul-Index")
// Test caching
{
// Fetch it again
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
issued2 := &structs.IssuedCert{}
require.NoError(t, dec.Decode(issued2))
require.Equal(t, issued, issued2)
}
// Set a new CA
ca2 := connect.TestCAConfigSet(t, a, nil)
// Issue a blocking query to ensure that the cert gets updated appropriately
{
resp := httptest.NewRecorder()
req, _ := http.NewRequest("GET", "/v1/agent/connect/ca/leaf/test?index="+index, nil)
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
issued2 := &structs.IssuedCert{}
require.NoError(t, dec.Decode(issued2))
require.NotEqual(t, issued.CertPEM, issued2.CertPEM)
require.NotEqual(t, issued.PrivateKeyPEM, issued2.PrivateKeyPEM)
// Verify that the cert is signed by the new CA
requireLeafValidUnderCA(t, issued2, ca2)
// Should not be a cache hit! The data was updated in response to the blocking
// query being made.
require.Equal(t, "MISS", resp.Header().Get("X-Cache"))
}
t.Run("test non-blocking queries update leaf cert", func(t *testing.T) {
resp := httptest.NewRecorder()
obj, err := a.srv.AgentConnectCALeafCert(resp, req)
require.NoError(t, err)
// Get the issued cert
issued, ok := obj.(*structs.IssuedCert)
assert.True(t, ok)
// Verify that the cert is signed by the CA
requireLeafValidUnderCA(t, issued, ca2)
// Issue a non blocking query to ensure that the cert gets updated appropriately
{
// Set a new CA
ca3 := connect.TestCAConfigSet(t, a, nil)
resp := httptest.NewRecorder()
req, err := http.NewRequest("GET", "/v1/agent/connect/ca/leaf/test", nil)
require.NoError(t, err)
obj, err = a.srv.AgentConnectCALeafCert(resp, req)
require.NoError(t, err)
issued2 := obj.(*structs.IssuedCert)
require.NotEqual(t, issued.CertPEM, issued2.CertPEM)
require.NotEqual(t, issued.PrivateKeyPEM, issued2.PrivateKeyPEM)
// Verify that the cert is signed by the new CA
requireLeafValidUnderCA(t, issued2, ca3)
// Should not be a cache hit!
require.Equal(t, "MISS", resp.Header().Get("X-Cache"))
}
// Test caching for the leaf cert
{
for fetched := 0; fetched < 4; fetched++ {
// Fetch it again
resp := httptest.NewRecorder()
obj2, err := a.srv.AgentConnectCALeafCert(resp, req)
require.NoError(t, err)
require.Equal(t, obj, obj2)
}
}
})
}
// Test we can request a leaf cert for a service we have permission for
// but is not local to this agent.
func TestAgentConnectCALeafCert_goodNotLocal(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := StartTestAgent(t, TestAgent{Overrides: `
connect {
test_ca_leaf_root_change_spread = "1ns"
}
`})
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
testrpc.WaitForActiveCARoot(t, a.RPC, "dc1", nil)
// CA already setup by default by NewTestAgent but force a new one so we can
// verify it was signed easily.
ca1 := connect.TestCAConfigSet(t, a, nil)
{
// Register a non-local service (central catalog)
args := &structs.RegisterRequest{
Node: "foo",
Address: "127.0.0.1",
Service: &structs.NodeService{
Service: "test",
Address: "127.0.0.1",
Port: 8080,
},
}
req, _ := http.NewRequest("PUT", "/v1/catalog/register", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if !assert.Equal(t, 200, resp.Code) {
t.Log("Body: ", resp.Body.String())
}
}
// List
req, _ := http.NewRequest("GET", "/v1/agent/connect/ca/leaf/test", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, "MISS", resp.Header().Get("X-Cache"))
// Get the issued cert
dec := json.NewDecoder(resp.Body)
issued := &structs.IssuedCert{}
require.NoError(t, dec.Decode(issued))
// Verify that the cert is signed by the CA
requireLeafValidUnderCA(t, issued, ca1)
// Verify blocking index
assert.True(t, issued.ModifyIndex > 0)
assert.Equal(t, fmt.Sprintf("%d", issued.ModifyIndex),
resp.Header().Get("X-Consul-Index"))
// Test caching
{
// Fetch it again
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
issued2 := &structs.IssuedCert{}
require.NoError(t, dec.Decode(issued2))
require.Equal(t, issued, issued2)
}
// Test Blocking - see https://github.com/hashicorp/consul/issues/4462
{
// Fetch it again
resp := httptest.NewRecorder()
blockingReq, _ := http.NewRequest("GET", fmt.Sprintf("/v1/agent/connect/ca/leaf/test?wait=125ms&index=%d", issued.ModifyIndex), nil)
doneCh := make(chan struct{})
go func() {
a.srv.h.ServeHTTP(resp, blockingReq)
close(doneCh)
}()
select {
case <-time.After(500 * time.Millisecond):
require.FailNow(t, "Shouldn't block for this long - not respecting wait parameter in the query")
case <-doneCh:
}
}
// Test that caching is updated in the background
{
// Set a new CA
ca := connect.TestCAConfigSet(t, a, nil)
retry.Run(t, func(r *retry.R) {
resp := httptest.NewRecorder()
// Try and sign again (note no index/wait arg since cache should update in
// background even if we aren't actively blocking)
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
issued2 := &structs.IssuedCert{}
require.NoError(t, dec.Decode(issued2))
if issued.CertPEM == issued2.CertPEM {
r.Fatalf("leaf has not updated")
}
// Got a new leaf. Sanity check it's a whole new key as well as different
// cert.
if issued.PrivateKeyPEM == issued2.PrivateKeyPEM {
r.Fatalf("new leaf has same private key as before")
}
// Verify that the cert is signed by the new CA
requireLeafValidUnderCA(t, issued2, ca)
require.NotEqual(t, issued, issued2)
})
}
}
func TestAgentConnectCALeafCert_Vault_doesNotChurnLeafCertsAtIdle(t *testing.T) {
ca.SkipIfVaultNotPresent(t)
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
testVault := ca.NewTestVaultServer(t)
defer testVault.Stop()
a := StartTestAgent(t, TestAgent{Overrides: fmt.Sprintf(`
connect {
test_ca_leaf_root_change_spread = "1ns"
ca_provider = "vault"
ca_config {
address = %[1]q
token = %[2]q
root_pki_path = "pki-root/"
intermediate_pki_path = "pki-intermediate/"
}
}
`, testVault.Addr, testVault.RootToken)})
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
testrpc.WaitForActiveCARoot(t, a.RPC, "dc1", nil)
var ca1 *structs.CARoot
{
args := &structs.DCSpecificRequest{Datacenter: "dc1"}
var reply structs.IndexedCARoots
require.NoError(t, a.RPC("ConnectCA.Roots", args, &reply))
for _, r := range reply.Roots {
if r.ID == reply.ActiveRootID {
ca1 = r
break
}
}
require.NotNil(t, ca1)
}
{
// Register a local service
args := &structs.ServiceDefinition{
ID: "foo",
Name: "test",
Address: "127.0.0.1",
Port: 8000,
Check: structs.CheckType{
TTL: 15 * time.Second,
},
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if !assert.Equal(t, 200, resp.Code) {
t.Log("Body: ", resp.Body.String())
}
}
// List
req, _ := http.NewRequest("GET", "/v1/agent/connect/ca/leaf/test", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, "MISS", resp.Header().Get("X-Cache"))
// Get the issued cert
dec := json.NewDecoder(resp.Body)
issued := &structs.IssuedCert{}
require.NoError(t, dec.Decode(issued))
// Verify that the cert is signed by the CA
requireLeafValidUnderCA(t, issued, ca1)
// Verify blocking index
assert.True(t, issued.ModifyIndex > 0)
assert.Equal(t, fmt.Sprintf("%d", issued.ModifyIndex),
resp.Header().Get("X-Consul-Index"))
// Test caching
{
// Fetch it again
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
issued2 := &structs.IssuedCert{}
require.NoError(t, dec.Decode(issued2))
require.Equal(t, issued, issued2)
}
// Test that we aren't churning leaves for no reason at idle.
{
ch := make(chan error, 1)
go func() {
req, _ := http.NewRequest("GET", "/v1/agent/connect/ca/leaf/test?index="+strconv.Itoa(int(issued.ModifyIndex)), nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
if resp.Code != http.StatusOK {
ch <- fmt.Errorf(resp.Body.String())
return
}
dec := json.NewDecoder(resp.Body)
issued2 := &structs.IssuedCert{}
if err := dec.Decode(issued2); err != nil {
ch <- err
} else {
if issued.CertPEM == issued2.CertPEM {
ch <- fmt.Errorf("leaf woke up unexpectedly with same cert")
} else {
ch <- fmt.Errorf("leaf woke up unexpectedly with new cert")
}
}
}()
start := time.Now()
select {
case <-time.After(5 * time.Second):
case err := <-ch:
dur := time.Since(start)
t.Fatalf("unexpected return from blocking query; leaf churned during idle period, took %s: %v", dur, err)
}
}
}
func TestAgentConnectCALeafCert_secondaryDC_good(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a1 := StartTestAgent(t, TestAgent{Name: "dc1", HCL: `
datacenter = "dc1"
primary_datacenter = "dc1"
`, Overrides: `
connect {
test_ca_leaf_root_change_spread = "1ns"
}
`})
defer a1.Shutdown()
testrpc.WaitForTestAgent(t, a1.RPC, "dc1")
a2 := StartTestAgent(t, TestAgent{Name: "dc2", HCL: `
datacenter = "dc2"
primary_datacenter = "dc1"
`, Overrides: `
connect {
test_ca_leaf_root_change_spread = "1ns"
}
`})
defer a2.Shutdown()
testrpc.WaitForTestAgent(t, a2.RPC, "dc2")
// Wait for the WAN join.
addr := fmt.Sprintf("127.0.0.1:%d", a1.Config.SerfPortWAN)
_, err := a2.JoinWAN([]string{addr})
require.NoError(t, err)
testrpc.WaitForLeader(t, a1.RPC, "dc1")
testrpc.WaitForLeader(t, a2.RPC, "dc2")
retry.Run(t, func(r *retry.R) {
if got, want := len(a1.WANMembers()), 2; got < want {
r.Fatalf("got %d WAN members want at least %d", got, want)
}
})
// CA already setup by default by NewTestAgent but force a new one so we can
// verify it was signed easily.
dc1_ca1 := connect.TestCAConfigSet(t, a1, nil)
// Wait until root is updated in both dcs.
waitForActiveCARoot(t, a1.srv, dc1_ca1)
waitForActiveCARoot(t, a2.srv, dc1_ca1)
{
// Register a local service in the SECONDARY
args := &structs.ServiceDefinition{
ID: "foo",
Name: "test",
Address: "127.0.0.1",
Port: 8000,
Check: structs.CheckType{
TTL: 15 * time.Second,
},
}
req, _ := http.NewRequest("PUT", "/v1/agent/service/register", jsonReader(args))
resp := httptest.NewRecorder()
a2.srv.h.ServeHTTP(resp, req)
if !assert.Equal(t, 200, resp.Code) {
t.Log("Body: ", resp.Body.String())
}
}
// List
req, err := http.NewRequest("GET", "/v1/agent/connect/ca/leaf/test", nil)
require.NoError(t, err)
resp := httptest.NewRecorder()
a2.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
require.Equal(t, "MISS", resp.Header().Get("X-Cache"))
// Get the issued cert
dec := json.NewDecoder(resp.Body)
issued := &structs.IssuedCert{}
require.NoError(t, dec.Decode(issued))
// Verify that the cert is signed by the CA
requireLeafValidUnderCA(t, issued, dc1_ca1)
// Verify blocking index
assert.True(t, issued.ModifyIndex > 0)
assert.Equal(t, fmt.Sprintf("%d", issued.ModifyIndex),
resp.Header().Get("X-Consul-Index"))
// Test caching
{
// Fetch it again
resp := httptest.NewRecorder()
a2.srv.h.ServeHTTP(resp, req)
dec := json.NewDecoder(resp.Body)
issued2 := &structs.IssuedCert{}
require.NoError(t, dec.Decode(issued2))
require.Equal(t, issued, issued2)
}
// Test that we aren't churning leaves for no reason at idle.
{
ch := make(chan error, 1)
go func() {
req, _ := http.NewRequest("GET", "/v1/agent/connect/ca/leaf/test?index="+strconv.Itoa(int(issued.ModifyIndex)), nil)
resp := httptest.NewRecorder()
a2.srv.h.ServeHTTP(resp, req)
if resp.Code != http.StatusOK {
ch <- fmt.Errorf(resp.Body.String())
return
}
dec := json.NewDecoder(resp.Body)
issued2 := &structs.IssuedCert{}
if err := dec.Decode(issued2); err != nil {
ch <- err
} else {
if issued.CertPEM == issued2.CertPEM {
ch <- fmt.Errorf("leaf woke up unexpectedly with same cert")
} else {
ch <- fmt.Errorf("leaf woke up unexpectedly with new cert")
}
}
}()
start := time.Now()
// Before applying the fix from PR-6513 this would reliably wake up
// after ~20ms with a new cert. Since this test is necessarily a bit
// timing dependent we'll chill out for 5 seconds which should be enough
// time to disprove the original bug.
select {
case <-time.After(5 * time.Second):
case err := <-ch:
dur := time.Since(start)
t.Fatalf("unexpected return from blocking query; leaf churned during idle period, took %s: %v", dur, err)
}
}
// Set a new CA
dc1_ca2 := connect.TestCAConfigSet(t, a2, nil)
// Wait until root is updated in both dcs.
waitForActiveCARoot(t, a1.srv, dc1_ca2)
waitForActiveCARoot(t, a2.srv, dc1_ca2)
// Test that caching is updated in the background
retry.Run(t, func(r *retry.R) {
resp := httptest.NewRecorder()
// Try and sign again (note no index/wait arg since cache should update in
// background even if we aren't actively blocking)
a2.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
dec := json.NewDecoder(resp.Body)
issued2 := &structs.IssuedCert{}
require.NoError(t, dec.Decode(issued2))
if issued.CertPEM == issued2.CertPEM {
r.Fatalf("leaf has not updated")
}
// Got a new leaf. Sanity check it's a whole new key as well as different
// cert.
if issued.PrivateKeyPEM == issued2.PrivateKeyPEM {
r.Fatalf("new leaf has same private key as before")
}
// Verify that the cert is signed by the new CA
requireLeafValidUnderCA(t, issued2, dc1_ca2)
require.NotEqual(t, issued, issued2)
})
}
func waitForActiveCARoot(t *testing.T, srv *HTTPHandlers, expect *structs.CARoot) {
retry.Run(t, func(r *retry.R) {
req, _ := http.NewRequest("GET", "/v1/agent/connect/ca/roots", nil)
resp := httptest.NewRecorder()
srv.h.ServeHTTP(resp, req)
if http.StatusOK != resp.Code {
t.Fatalf("expected 200 but got %v", resp.Code)
}
dec := json.NewDecoder(resp.Body)
roots := &structs.IndexedCARoots{}
require.NoError(t, dec.Decode(roots))
var root *structs.CARoot
for _, r := range roots.Roots {
if r.ID == roots.ActiveRootID {
root = r
break
}
}
if root == nil {
r.Fatal("no active root")
}
if root.ID != expect.ID {
r.Fatalf("current active root is %s; waiting for %s", root.ID, expect.ID)
}
})
}
func requireLeafValidUnderCA(t *testing.T, issued *structs.IssuedCert, ca *structs.CARoot) {
leaf, intermediates, err := connect.ParseLeafCerts(issued.CertPEM)
require.NoError(t, err)
roots := x509.NewCertPool()
require.True(t, roots.AppendCertsFromPEM([]byte(ca.RootCert)))
_, err = leaf.Verify(x509.VerifyOptions{
Roots: roots,
Intermediates: intermediates,
})
require.NoError(t, err)
// Verify the private key matches. tls.LoadX509Keypair does this for us!
_, err = tls.X509KeyPair([]byte(issued.CertPEM), []byte(issued.PrivateKeyPEM))
require.NoError(t, err)
}
func TestAgentConnectAuthorize_badBody(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
args := []string{}
req, _ := http.NewRequest("POST", "/v1/agent/connect/authorize", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusBadRequest, resp.Code)
assert.Contains(t, resp.Body.String(), "decode failed")
}
func TestAgentConnectAuthorize_noTarget(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
args := &structs.ConnectAuthorizeRequest{}
req, _ := http.NewRequest("POST", "/v1/agent/connect/authorize", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusBadRequest, resp.Code)
assert.Contains(t, resp.Body.String(), "Target service must be specified")
}
// Client ID is not in the valid URI format
func TestAgentConnectAuthorize_idInvalidFormat(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
args := &structs.ConnectAuthorizeRequest{
Target: "web",
ClientCertURI: "tubes",
}
req, _ := http.NewRequest("POST", "/v1/agent/connect/authorize", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusBadRequest, resp.Code)
assert.Contains(t, resp.Body.String(), "ClientCertURI not a valid Connect identifier")
}
// Client ID is a valid URI but its not a service URI
func TestAgentConnectAuthorize_idNotService(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
args := &structs.ConnectAuthorizeRequest{
Target: "web",
ClientCertURI: "spiffe://1234.consul",
}
req, _ := http.NewRequest("POST", "/v1/agent/connect/authorize", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusBadRequest, resp.Code)
assert.Contains(t, resp.Body.String(), "ClientCertURI not a valid Service identifier")
}
// Test when there is an intention allowing the connection
func TestAgentConnectAuthorize_allow(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
target := "db"
// Create some intentions
var ixnId string
{
req := structs.IntentionRequest{
Datacenter: "dc1",
Op: structs.IntentionOpCreate,
Intention: structs.TestIntention(t),
}
req.Intention.SourceNS = structs.IntentionDefaultNamespace
req.Intention.SourceName = "web"
req.Intention.DestinationNS = structs.IntentionDefaultNamespace
req.Intention.DestinationName = target
req.Intention.Action = structs.IntentionActionAllow
require.Nil(t, a.RPC("Intention.Apply", &req, &ixnId))
}
args := &structs.ConnectAuthorizeRequest{
Target: target,
ClientCertURI: connect.TestSpiffeIDService(t, "web").URI().String(),
}
req, _ := http.NewRequest("POST", "/v1/agent/connect/authorize", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, 200, resp.Code)
require.Equal(t, "MISS", resp.Header().Get("X-Cache"))
dec := json.NewDecoder(resp.Body)
obj := &connectAuthorizeResp{}
require.NoError(t, dec.Decode(obj))
require.True(t, obj.Authorized)
require.Contains(t, obj.Reason, "Matched")
// Make the request again
{
req, _ := http.NewRequest("POST", "/v1/agent/connect/authorize", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, 200, resp.Code)
dec := json.NewDecoder(resp.Body)
obj := &connectAuthorizeResp{}
require.NoError(t, dec.Decode(obj))
require.True(t, obj.Authorized)
require.Contains(t, obj.Reason, "Matched")
// That should've been a cache hit.
require.Equal(t, "HIT", resp.Header().Get("X-Cache"))
}
// Change the intention
{
req := structs.IntentionRequest{
Datacenter: "dc1",
Op: structs.IntentionOpUpdate,
Intention: structs.TestIntention(t),
}
req.Intention.ID = ixnId
req.Intention.SourceNS = structs.IntentionDefaultNamespace
req.Intention.SourceName = "web"
req.Intention.DestinationNS = structs.IntentionDefaultNamespace
req.Intention.DestinationName = target
req.Intention.Action = structs.IntentionActionDeny
require.Nil(t, a.RPC("Intention.Apply", &req, &ixnId))
}
// Short sleep lets the cache background refresh happen
time.Sleep(100 * time.Millisecond)
// Make the request again
{
req, _ := http.NewRequest("POST", "/v1/agent/connect/authorize", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, 200, resp.Code)
dec := json.NewDecoder(resp.Body)
obj := &connectAuthorizeResp{}
require.NoError(t, dec.Decode(obj))
require.False(t, obj.Authorized)
require.Contains(t, obj.Reason, "Matched")
// That should've been a cache hit, too, since it updated in the
// background.
require.Equal(t, "HIT", resp.Header().Get("X-Cache"))
}
}
// Test when there is an intention denying the connection
func TestAgentConnectAuthorize_deny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
target := "db"
// Create some intentions
{
req := structs.IntentionRequest{
Datacenter: "dc1",
Op: structs.IntentionOpCreate,
Intention: structs.TestIntention(t),
}
req.Intention.SourceNS = structs.IntentionDefaultNamespace
req.Intention.SourceName = "web"
req.Intention.DestinationNS = structs.IntentionDefaultNamespace
req.Intention.DestinationName = target
req.Intention.Action = structs.IntentionActionDeny
var reply string
assert.Nil(t, a.RPC("Intention.Apply", &req, &reply))
}
args := &structs.ConnectAuthorizeRequest{
Target: target,
ClientCertURI: connect.TestSpiffeIDService(t, "web").URI().String(),
}
req, _ := http.NewRequest("POST", "/v1/agent/connect/authorize", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
assert.Equal(t, 200, resp.Code)
dec := json.NewDecoder(resp.Body)
obj := &connectAuthorizeResp{}
require.NoError(t, dec.Decode(obj))
assert.False(t, obj.Authorized)
assert.Contains(t, obj.Reason, "Matched")
}
// Test when there is an intention allowing service with a different trust
// domain. We allow this because migration between trust domains shouldn't cause
// an outage even if we have stale info about current trusted domains. It's safe
// because the CA root is either unique to this cluster and not used to sign
// anything external, or path validation can be used to ensure that the CA can
// only issue certs that are valid for the specific cluster trust domain at x509
// level which is enforced by TLS handshake.
func TestAgentConnectAuthorize_allowTrustDomain(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
target := "db"
// Create some intentions
{
req := structs.IntentionRequest{
Datacenter: "dc1",
Op: structs.IntentionOpCreate,
Intention: structs.TestIntention(t),
}
req.Intention.SourceNS = structs.IntentionDefaultNamespace
req.Intention.SourceName = "web"
req.Intention.DestinationNS = structs.IntentionDefaultNamespace
req.Intention.DestinationName = target
req.Intention.Action = structs.IntentionActionAllow
var reply string
require.NoError(t, a.RPC("Intention.Apply", &req, &reply))
}
{
args := &structs.ConnectAuthorizeRequest{
Target: target,
ClientCertURI: "spiffe://fake-domain.consul/ns/default/dc/dc1/svc/web",
}
req, _ := http.NewRequest("POST", "/v1/agent/connect/authorize", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
assert.Equal(t, 200, resp.Code)
dec := json.NewDecoder(resp.Body)
obj := &connectAuthorizeResp{}
require.NoError(t, dec.Decode(obj))
require.True(t, obj.Authorized)
require.Contains(t, obj.Reason, "Matched")
}
}
func TestAgentConnectAuthorize_denyWildcard(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
target := "db"
// Create some intentions
{
// Deny wildcard to DB
req := structs.IntentionRequest{
Datacenter: "dc1",
Op: structs.IntentionOpCreate,
Intention: structs.TestIntention(t),
}
req.Intention.SourceNS = structs.IntentionDefaultNamespace
req.Intention.SourceName = "*"
req.Intention.DestinationNS = structs.IntentionDefaultNamespace
req.Intention.DestinationName = target
req.Intention.Action = structs.IntentionActionDeny
var reply string
require.NoError(t, a.RPC("Intention.Apply", &req, &reply))
}
{
// Allow web to DB
req := structs.IntentionRequest{
Datacenter: "dc1",
Op: structs.IntentionOpCreate,
Intention: structs.TestIntention(t),
}
req.Intention.SourceNS = structs.IntentionDefaultNamespace
req.Intention.SourceName = "web"
req.Intention.DestinationNS = structs.IntentionDefaultNamespace
req.Intention.DestinationName = target
req.Intention.Action = structs.IntentionActionAllow
var reply string
assert.Nil(t, a.RPC("Intention.Apply", &req, &reply))
}
// Web should be allowed
{
args := &structs.ConnectAuthorizeRequest{
Target: target,
ClientCertURI: connect.TestSpiffeIDService(t, "web").URI().String(),
}
req, _ := http.NewRequest("POST", "/v1/agent/connect/authorize", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
assert.Equal(t, 200, resp.Code)
dec := json.NewDecoder(resp.Body)
obj := &connectAuthorizeResp{}
require.NoError(t, dec.Decode(obj))
assert.True(t, obj.Authorized)
assert.Contains(t, obj.Reason, "Matched")
}
// API should be denied
{
args := &structs.ConnectAuthorizeRequest{
Target: target,
ClientCertURI: connect.TestSpiffeIDService(t, "api").URI().String(),
}
req, _ := http.NewRequest("POST", "/v1/agent/connect/authorize", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
assert.Equal(t, 200, resp.Code)
dec := json.NewDecoder(resp.Body)
obj := &connectAuthorizeResp{}
require.NoError(t, dec.Decode(obj))
assert.False(t, obj.Authorized)
assert.Contains(t, obj.Reason, "Matched")
}
}
// Test that authorize fails without service:write for the target service.
func TestAgentConnectAuthorize_serviceWrite(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
token := createACLTokenWithServicePolicy(t, a.srv, "read")
args := &structs.ConnectAuthorizeRequest{
Target: "test",
ClientCertURI: connect.TestSpiffeIDService(t, "web").URI().String(),
}
req, _ := http.NewRequest("POST",
"/v1/agent/connect/authorize?token="+token, jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
assert.Equal(t, http.StatusForbidden, resp.Code)
}
// Test when no intentions match w/ a default deny policy
func TestAgentConnectAuthorize_defaultDeny(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, TestACLConfig())
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
args := &structs.ConnectAuthorizeRequest{
Target: "foo",
ClientCertURI: connect.TestSpiffeIDService(t, "web").URI().String(),
}
req, _ := http.NewRequest("POST", "/v1/agent/connect/authorize?token=root", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
assert.Equal(t, 200, resp.Code)
dec := json.NewDecoder(resp.Body)
obj := &connectAuthorizeResp{}
require.NoError(t, dec.Decode(obj))
assert.False(t, obj.Authorized)
assert.Contains(t, obj.Reason, "Default behavior")
}
// Test when no intentions match w/ a default allow policy
func TestAgentConnectAuthorize_defaultAllow(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
dc1 := "dc1"
a := NewTestAgent(t, `
primary_datacenter = "`+dc1+`"
acl {
enabled = true
default_policy = "allow"
tokens {
initial_management = "root"
agent = "root"
agent_recovery = "towel"
}
}
`)
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, dc1)
args := &structs.ConnectAuthorizeRequest{
Target: "foo",
ClientCertURI: connect.TestSpiffeIDService(t, "web").URI().String(),
}
req, _ := http.NewRequest("POST", "/v1/agent/connect/authorize?token=root", jsonReader(args))
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
assert.Equal(t, 200, resp.Code)
dec := json.NewDecoder(resp.Body)
obj := &connectAuthorizeResp{}
require.NoError(t, dec.Decode(obj))
assert.True(t, obj.Authorized)
assert.Contains(t, obj.Reason, "Default behavior")
}
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
func TestAgent_Host(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
t.Parallel()
dc1 := "dc1"
a := NewTestAgent(t, `
primary_datacenter = "`+dc1+`"
acl {
enabled = true
default_policy = "allow"
tokens {
initial_management = "initial-management"
agent = "agent"
agent_recovery = "towel"
}
}
`)
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
req, _ := http.NewRequest("GET", "/v1/agent/host?token=initial-management", nil)
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
resp := httptest.NewRecorder()
// TODO: AgentHost should write to response so that we can test using ServeHTTP()
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
respRaw, err := a.srv.AgentHost(resp, req)
assert.Nil(t, err)
assert.Equal(t, http.StatusOK, resp.Code)
assert.NotNil(t, respRaw)
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
obj := respRaw.(*debug.HostInfo)
assert.NotNil(t, obj.CollectionTime)
assert.Empty(t, obj.Errors)
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
}
func TestAgent_HostBadACL(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
t.Parallel()
dc1 := "dc1"
a := NewTestAgent(t, `
primary_datacenter = "`+dc1+`"
acl {
enabled = true
default_policy = "deny"
tokens {
initial_management = "root"
agent = "agent"
agent_recovery = "towel"
}
}
`)
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
defer a.Shutdown()
testrpc.WaitForLeader(t, a.RPC, "dc1")
req, _ := http.NewRequest("GET", "/v1/agent/host?token=agent", nil)
resp := httptest.NewRecorder()
// TODO: AgentHost should write to response so that we can test using ServeHTTP()
_, err := a.srv.AgentHost(resp, req)
assert.EqualError(t, err, "ACL not found")
assert.Equal(t, http.StatusOK, resp.Code)
New command: consul debug (#4754) * agent/debug: add package for debugging, host info * api: add v1/agent/host endpoint * agent: add v1/agent/host endpoint * command/debug: implementation of static capture * command/debug: tests and only configured targets * agent/debug: add basic test for host metrics * command/debug: add methods for dynamic data capture * api: add debug/pprof endpoints * command/debug: add pprof * command/debug: timing, wg, logs to disk * vendor: add gopsutil/disk * command/debug: add a usage section * website: add docs for consul debug * agent/host: require operator:read * api/host: improve docs and no retry timing * command/debug: fail on extra arguments * command/debug: fixup file permissions to 0644 * command/debug: remove server flags * command/debug: improve clarity of usage section * api/debug: add Trace for profiling, fix profile * command/debug: capture profile and trace at the same time * command/debug: add index document * command/debug: use "clusters" in place of members * command/debug: remove address in output * command/debug: improve comment on metrics sleep * command/debug: clarify usage * agent: always register pprof handlers and protect This will allow us to avoid a restart of a target agent for profiling by always registering the pprof handlers. Given this is a potentially sensitive path, it is protected with an operator:read ACL and enable debug being set to true on the target agent. enable_debug still requires a restart. If ACLs are disabled, enable_debug is sufficient. * command/debug: use trace.out instead of .prof More in line with golang docs. * agent: fix comment wording * agent: wrap table driven tests in t.run()
2018-10-17 20:20:35 +00:00
}
// Thie tests that a proxy with an ExposeConfig is returned as expected.
func TestAgent_Services_ExposeConfig(t *testing.T) {
if testing.Short() {
t.Skip("too slow for testing.Short")
}
t.Parallel()
a := NewTestAgent(t, "")
defer a.Shutdown()
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
srv1 := &structs.NodeService{
Kind: structs.ServiceKindConnectProxy,
ID: "proxy-id",
Service: "proxy-name",
Port: 8443,
Proxy: structs.ConnectProxyConfig{
Expose: structs.ExposeConfig{
Checks: true,
Paths: []structs.ExposePath{
{
ListenerPort: 8080,
LocalPathPort: 21500,
Protocol: "http2",
Path: "/metrics",
},
},
},
},
}
a.State.AddService(srv1, "")
req, _ := http.NewRequest("GET", "/v1/agent/services", nil)
resp := httptest.NewRecorder()
a.srv.h.ServeHTTP(resp, req)
require.Equal(t, http.StatusOK, resp.Code)
dec := json.NewDecoder(resp.Body)
val := make(map[string]*api.AgentService)
require.NoError(t, dec.Decode(&val))
require.Len(t, val, 1)
actual := val["proxy-id"]
require.NotNil(t, actual)
require.Equal(t, api.ServiceKindConnectProxy, actual.Kind)
// Proxy.ToAPI() creates an empty Upstream list instead of keeping nil so do the same with actual.
if actual.Proxy.Upstreams == nil {
actual.Proxy.Upstreams = make([]api.Upstream, 0)
}
require.Equal(t, srv1.Proxy.ToAPI(), actual.Proxy)
}