open-consul/agent/consul/fsm/snapshot_oss_test.go

748 lines
19 KiB
Go
Raw Normal View History

package fsm
import (
"bytes"
"reflect"
"testing"
"time"
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
"github.com/hashicorp/consul/acl"
"github.com/hashicorp/consul/agent/connect"
2017-12-12 00:38:52 +00:00
"github.com/hashicorp/consul/agent/consul/autopilot"
"github.com/hashicorp/consul/agent/consul/state"
"github.com/hashicorp/consul/agent/structs"
"github.com/hashicorp/consul/api"
"github.com/hashicorp/consul/lib"
"github.com/hashicorp/consul/sdk/testutil"
"github.com/hashicorp/go-raftchunking"
"github.com/stretchr/testify/assert"
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
"github.com/stretchr/testify/require"
)
func TestFSM_SnapshotRestore_OSS(t *testing.T) {
t.Parallel()
assert := assert.New(t)
require := require.New(t)
logger := testutil.Logger(t)
fsm, err := New(nil, logger)
if err != nil {
t.Fatalf("err: %v", err)
}
// Add some state
node1 := &structs.Node{
ID: "610918a6-464f-fa9b-1a95-03bd6e88ed92",
Node: "foo",
Datacenter: "dc1",
Address: "127.0.0.1",
}
node2 := &structs.Node{
ID: "40e4a748-2192-161a-0510-9bf59fe950b5",
Node: "baz",
Datacenter: "dc1",
Address: "127.0.0.2",
TaggedAddresses: map[string]string{
"hello": "1.2.3.4",
},
Meta: map[string]string{
"testMeta": "testing123",
},
}
require.NoError(fsm.state.EnsureNode(1, node1))
require.NoError(fsm.state.EnsureNode(2, node2))
// Add a service instance with Connect config.
connectConf := structs.ServiceConnect{
Native: true,
}
fsm.state.EnsureService(3, "foo", &structs.NodeService{
ID: "web",
Service: "web",
Tags: nil,
Address: "127.0.0.1",
Port: 80,
Connect: connectConf,
})
fsm.state.EnsureService(4, "foo", &structs.NodeService{ID: "db", Service: "db", Tags: []string{"primary"}, Address: "127.0.0.1", Port: 5000})
fsm.state.EnsureService(5, "baz", &structs.NodeService{ID: "web", Service: "web", Tags: nil, Address: "127.0.0.2", Port: 80})
fsm.state.EnsureService(6, "baz", &structs.NodeService{ID: "db", Service: "db", Tags: []string{"secondary"}, Address: "127.0.0.2", Port: 5000})
fsm.state.EnsureCheck(7, &structs.HealthCheck{
Node: "foo",
CheckID: "web",
Name: "web connectivity",
Status: api.HealthPassing,
ServiceID: "web",
})
fsm.state.KVSSet(8, &structs.DirEntry{
Key: "/test",
Value: []byte("foo"),
})
session := &structs.Session{ID: generateUUID(), Node: "foo"}
fsm.state.SessionCreate(9, session)
policy := &structs.ACLPolicy{
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
ID: structs.ACLPolicyGlobalManagementID,
Name: "global-management",
Description: "Builtin Policy that grants unlimited access",
Rules: structs.ACLPolicyGlobalManagement,
Syntax: acl.SyntaxCurrent,
}
policy.SetHash(true)
require.NoError(fsm.state.ACLPolicySet(1, policy))
role := &structs.ACLRole{
ID: "86dedd19-8fae-4594-8294-4e6948a81f9a",
Name: "some-role",
Description: "test snapshot role",
ServiceIdentities: []*structs.ACLServiceIdentity{
&structs.ACLServiceIdentity{
ServiceName: "example",
},
},
}
role.SetHash(true)
require.NoError(fsm.state.ACLRoleSet(1, role))
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
token := &structs.ACLToken{
AccessorID: "30fca056-9fbb-4455-b94a-bf0e2bc575d6",
SecretID: "cbe1c6fd-d865-4034-9d6d-64fef7fb46a9",
Description: "Bootstrap Token (Global Management)",
Policies: []structs.ACLTokenPolicyLink{
{
ID: structs.ACLPolicyGlobalManagementID,
},
},
CreateTime: time.Now(),
Local: false,
// DEPRECATED (ACL-Legacy-Compat) - This is used so that the bootstrap token is still visible via the v1 acl APIs
Type: structs.ACLTokenTypeManagement,
}
require.NoError(fsm.state.ACLBootstrap(10, 0, token, false))
method := &structs.ACLAuthMethod{
Name: "some-method",
Type: "testing",
Description: "test snapshot auth method",
Config: map[string]interface{}{
"SessionID": "952ebfa8-2a42-46f0-bcd3-fd98a842000e",
},
}
require.NoError(fsm.state.ACLAuthMethodSet(1, method))
bindingRule := &structs.ACLBindingRule{
ID: "85184c52-5997-4a84-9817-5945f2632a17",
Description: "test snapshot binding rule",
AuthMethod: "some-method",
Selector: "serviceaccount.namespace==default",
BindType: structs.BindingRuleBindTypeService,
BindName: "${serviceaccount.name}",
}
require.NoError(fsm.state.ACLBindingRuleSet(1, bindingRule))
fsm.state.KVSSet(11, &structs.DirEntry{
Key: "/remove",
Value: []byte("foo"),
})
fsm.state.KVSDelete(12, "/remove", nil)
idx, _, err := fsm.state.KVSList(nil, "/remove", nil)
if err != nil {
t.Fatalf("err: %s", err)
}
if idx != 12 {
t.Fatalf("bad index: %d", idx)
}
updates := structs.Coordinates{
&structs.Coordinate{
Node: "baz",
Coord: generateRandomCoordinate(),
},
&structs.Coordinate{
Node: "foo",
Coord: generateRandomCoordinate(),
},
}
if err := fsm.state.CoordinateBatchUpdate(13, updates); err != nil {
t.Fatalf("err: %s", err)
}
query := structs.PreparedQuery{
ID: generateUUID(),
Service: structs.ServiceQuery{
Service: "web",
},
RaftIndex: structs.RaftIndex{
CreateIndex: 14,
ModifyIndex: 14,
},
}
if err := fsm.state.PreparedQuerySet(14, &query); err != nil {
t.Fatalf("err: %s", err)
}
2017-12-12 00:38:52 +00:00
autopilotConf := &autopilot.Config{
CleanupDeadServers: true,
LastContactThreshold: 100 * time.Millisecond,
MaxTrailingLogs: 222,
}
if err := fsm.state.AutopilotSetConfig(15, autopilotConf); err != nil {
t.Fatalf("err: %s", err)
}
// Intentions
ixn := structs.TestIntention(t)
ixn.ID = generateUUID()
ixn.RaftIndex = structs.RaftIndex{
CreateIndex: 14,
ModifyIndex: 14,
}
require.NoError(fsm.state.IntentionSet(14, ixn))
// CA Roots
roots := []*structs.CARoot{
connect.TestCA(t, nil),
connect.TestCA(t, nil),
}
for _, r := range roots[1:] {
r.Active = false
}
ok, err := fsm.state.CARootSetCAS(15, 0, roots)
require.NoError(err)
assert.True(ok)
ok, err = fsm.state.CASetProviderState(16, &structs.CAConsulProviderState{
ID: "asdf",
PrivateKey: "foo",
RootCert: "bar",
})
require.NoError(err)
assert.True(ok)
// CA Config
caConfig := &structs.CAConfiguration{
ClusterID: "foo",
Provider: "consul",
Config: map[string]interface{}{
"foo": "asdf",
"bar": 6.5,
},
}
err = fsm.state.CASetConfig(17, caConfig)
require.NoError(err)
// Config entries
serviceConfig := &structs.ServiceConfigEntry{
Kind: structs.ServiceDefaults,
Name: "foo",
Protocol: "http",
}
proxyConfig := &structs.ProxyConfigEntry{
Kind: structs.ProxyDefaults,
Name: "global",
}
require.NoError(fsm.state.EnsureConfigEntry(18, serviceConfig, structs.DefaultEnterpriseMeta()))
require.NoError(fsm.state.EnsureConfigEntry(19, proxyConfig, structs.DefaultEnterpriseMeta()))
ingress := &structs.IngressGatewayConfigEntry{
Kind: structs.IngressGateway,
Name: "ingress",
Listeners: []structs.IngressListener{
{
Port: 8080,
Protocol: "http",
Services: []structs.IngressService{
{
Name: "foo",
},
},
},
},
}
require.NoError(fsm.state.EnsureConfigEntry(20, ingress, structs.DefaultEnterpriseMeta()))
_, gatewayServices, err := fsm.state.GatewayServices(nil, "ingress", structs.DefaultEnterpriseMeta())
require.NoError(err)
wan federation via mesh gateways (#6884) This is like a Möbius strip of code due to the fact that low-level components (serf/memberlist) are connected to high-level components (the catalog and mesh-gateways) in a twisty maze of references which make it hard to dive into. With that in mind here's a high level summary of what you'll find in the patch: There are several distinct chunks of code that are affected: * new flags and config options for the server * retry join WAN is slightly different * retry join code is shared to discover primary mesh gateways from secondary datacenters * because retry join logic runs in the *agent* and the results of that operation for primary mesh gateways are needed in the *server* there are some methods like `RefreshPrimaryGatewayFallbackAddresses` that must occur at multiple layers of abstraction just to pass the data down to the right layer. * new cache type `FederationStateListMeshGatewaysName` for use in `proxycfg/xds` layers * the function signature for RPC dialing picked up a new required field (the node name of the destination) * several new RPCs for manipulating a FederationState object: `FederationState:{Apply,Get,List,ListMeshGateways}` * 3 read-only internal APIs for debugging use to invoke those RPCs from curl * raft and fsm changes to persist these FederationStates * replication for FederationStates as they are canonically stored in the Primary and replicated to the Secondaries. * a special derivative of anti-entropy that runs in secondaries to snapshot their local mesh gateway `CheckServiceNodes` and sync them into their upstream FederationState in the primary (this works in conjunction with the replication to distribute addresses for all mesh gateways in all DCs to all other DCs) * a "gateway locator" convenience object to make use of this data to choose the addresses of gateways to use for any given RPC or gossip operation to a remote DC. This gets data from the "retry join" logic in the agent and also directly calls into the FSM. * RPC (`:8300`) on the server sniffs the first byte of a new connection to determine if it's actually doing native TLS. If so it checks the ALPN header for protocol determination (just like how the existing system uses the type-byte marker). * 2 new kinds of protocols are exclusively decoded via this native TLS mechanism: one for ferrying "packet" operations (udp-like) from the gossip layer and one for "stream" operations (tcp-like). The packet operations re-use sockets (using length-prefixing) to cut down on TLS re-negotiation overhead. * the server instances specially wrap the `memberlist.NetTransport` when running with gateway federation enabled (in a `wanfed.Transport`). The general gist is that if it tries to dial a node in the SAME datacenter (deduced by looking at the suffix of the node name) there is no change. If dialing a DIFFERENT datacenter it is wrapped up in a TLS+ALPN blob and sent through some mesh gateways to eventually end up in a server's :8300 port. * a new flag when launching a mesh gateway via `consul connect envoy` to indicate that the servers are to be exposed. This sets a special service meta when registering the gateway into the catalog. * `proxycfg/xds` notice this metadata blob to activate additional watches for the FederationState objects as well as the location of all of the consul servers in that datacenter. * `xds:` if the extra metadata is in place additional clusters are defined in a DC to bulk sink all traffic to another DC's gateways. For the current datacenter we listen on a wildcard name (`server.<dc>.consul`) that load balances all servers as well as one mini-cluster per node (`<node>.server.<dc>.consul`) * the `consul tls cert create` command got a new flag (`-node`) to help create an additional SAN in certs that can be used with this flavor of federation.
2020-03-09 20:59:02 +00:00
// Raft Chunking
chunkState := &raftchunking.State{
ChunkMap: make(raftchunking.ChunkMap),
}
chunkState.ChunkMap[0] = []*raftchunking.ChunkInfo{
{
OpNum: 0,
SequenceNum: 0,
NumChunks: 3,
Data: []byte("foo"),
},
nil,
{
OpNum: 0,
SequenceNum: 2,
NumChunks: 3,
Data: []byte("bar"),
},
}
chunkState.ChunkMap[20] = []*raftchunking.ChunkInfo{
nil,
{
OpNum: 20,
SequenceNum: 1,
NumChunks: 2,
Data: []byte("bar"),
},
}
err = fsm.chunker.RestoreState(chunkState)
require.NoError(err)
wan federation via mesh gateways (#6884) This is like a Möbius strip of code due to the fact that low-level components (serf/memberlist) are connected to high-level components (the catalog and mesh-gateways) in a twisty maze of references which make it hard to dive into. With that in mind here's a high level summary of what you'll find in the patch: There are several distinct chunks of code that are affected: * new flags and config options for the server * retry join WAN is slightly different * retry join code is shared to discover primary mesh gateways from secondary datacenters * because retry join logic runs in the *agent* and the results of that operation for primary mesh gateways are needed in the *server* there are some methods like `RefreshPrimaryGatewayFallbackAddresses` that must occur at multiple layers of abstraction just to pass the data down to the right layer. * new cache type `FederationStateListMeshGatewaysName` for use in `proxycfg/xds` layers * the function signature for RPC dialing picked up a new required field (the node name of the destination) * several new RPCs for manipulating a FederationState object: `FederationState:{Apply,Get,List,ListMeshGateways}` * 3 read-only internal APIs for debugging use to invoke those RPCs from curl * raft and fsm changes to persist these FederationStates * replication for FederationStates as they are canonically stored in the Primary and replicated to the Secondaries. * a special derivative of anti-entropy that runs in secondaries to snapshot their local mesh gateway `CheckServiceNodes` and sync them into their upstream FederationState in the primary (this works in conjunction with the replication to distribute addresses for all mesh gateways in all DCs to all other DCs) * a "gateway locator" convenience object to make use of this data to choose the addresses of gateways to use for any given RPC or gossip operation to a remote DC. This gets data from the "retry join" logic in the agent and also directly calls into the FSM. * RPC (`:8300`) on the server sniffs the first byte of a new connection to determine if it's actually doing native TLS. If so it checks the ALPN header for protocol determination (just like how the existing system uses the type-byte marker). * 2 new kinds of protocols are exclusively decoded via this native TLS mechanism: one for ferrying "packet" operations (udp-like) from the gossip layer and one for "stream" operations (tcp-like). The packet operations re-use sockets (using length-prefixing) to cut down on TLS re-negotiation overhead. * the server instances specially wrap the `memberlist.NetTransport` when running with gateway federation enabled (in a `wanfed.Transport`). The general gist is that if it tries to dial a node in the SAME datacenter (deduced by looking at the suffix of the node name) there is no change. If dialing a DIFFERENT datacenter it is wrapped up in a TLS+ALPN blob and sent through some mesh gateways to eventually end up in a server's :8300 port. * a new flag when launching a mesh gateway via `consul connect envoy` to indicate that the servers are to be exposed. This sets a special service meta when registering the gateway into the catalog. * `proxycfg/xds` notice this metadata blob to activate additional watches for the FederationState objects as well as the location of all of the consul servers in that datacenter. * `xds:` if the extra metadata is in place additional clusters are defined in a DC to bulk sink all traffic to another DC's gateways. For the current datacenter we listen on a wildcard name (`server.<dc>.consul`) that load balances all servers as well as one mini-cluster per node (`<node>.server.<dc>.consul`) * the `consul tls cert create` command got a new flag (`-node`) to help create an additional SAN in certs that can be used with this flavor of federation.
2020-03-09 20:59:02 +00:00
// Federation states
fedState1 := &structs.FederationState{
Datacenter: "dc1",
MeshGateways: []structs.CheckServiceNode{
{
Node: &structs.Node{
ID: "664bac9f-4de7-4f1b-ad35-0e5365e8f329",
Node: "gateway1",
Datacenter: "dc1",
Address: "1.2.3.4",
},
Service: &structs.NodeService{
ID: "mesh-gateway",
Service: "mesh-gateway",
Kind: structs.ServiceKindMeshGateway,
Port: 1111,
Meta: map[string]string{structs.MetaWANFederationKey: "1"},
},
Checks: []*structs.HealthCheck{
{
Name: "web connectivity",
Status: api.HealthPassing,
ServiceID: "mesh-gateway",
},
},
},
{
Node: &structs.Node{
ID: "3fb9a696-8209-4eee-a1f7-48600deb9716",
Node: "gateway2",
Datacenter: "dc1",
Address: "9.8.7.6",
},
Service: &structs.NodeService{
ID: "mesh-gateway",
Service: "mesh-gateway",
Kind: structs.ServiceKindMeshGateway,
Port: 2222,
Meta: map[string]string{structs.MetaWANFederationKey: "1"},
},
Checks: []*structs.HealthCheck{
{
Name: "web connectivity",
Status: api.HealthPassing,
ServiceID: "mesh-gateway",
},
},
},
},
UpdatedAt: time.Now().UTC(),
}
fedState2 := &structs.FederationState{
Datacenter: "dc2",
MeshGateways: []structs.CheckServiceNode{
{
Node: &structs.Node{
ID: "0f92b02e-9f51-4aa2-861b-4ddbc3492724",
Node: "gateway1",
Datacenter: "dc2",
Address: "8.8.8.8",
},
Service: &structs.NodeService{
ID: "mesh-gateway",
Service: "mesh-gateway",
Kind: structs.ServiceKindMeshGateway,
Port: 3333,
Meta: map[string]string{structs.MetaWANFederationKey: "1"},
},
Checks: []*structs.HealthCheck{
{
Name: "web connectivity",
Status: api.HealthPassing,
ServiceID: "mesh-gateway",
},
},
},
{
Node: &structs.Node{
ID: "99a76121-1c3f-4023-88ef-805248beb10b",
Node: "gateway2",
Datacenter: "dc2",
Address: "5.5.5.5",
},
Service: &structs.NodeService{
ID: "mesh-gateway",
Service: "mesh-gateway",
Kind: structs.ServiceKindMeshGateway,
Port: 4444,
Meta: map[string]string{structs.MetaWANFederationKey: "1"},
},
Checks: []*structs.HealthCheck{
{
Name: "web connectivity",
Status: api.HealthPassing,
ServiceID: "mesh-gateway",
},
},
},
},
UpdatedAt: time.Now().UTC(),
}
require.NoError(fsm.state.FederationStateSet(21, fedState1))
require.NoError(fsm.state.FederationStateSet(22, fedState2))
// Snapshot
snap, err := fsm.Snapshot()
if err != nil {
t.Fatalf("err: %v", err)
}
defer snap.Release()
// Persist
buf := bytes.NewBuffer(nil)
sink := &MockSink{buf, false}
if err := snap.Persist(sink); err != nil {
t.Fatalf("err: %v", err)
}
// Try to restore on a new FSM
fsm2, err := New(nil, logger)
if err != nil {
t.Fatalf("err: %v", err)
}
// Do a restore
if err := fsm2.Restore(sink); err != nil {
t.Fatalf("err: %v", err)
}
// Verify the contents
_, nodes, err := fsm2.state.Nodes(nil)
if err != nil {
t.Fatalf("err: %s", err)
}
if len(nodes) != 2 {
t.Fatalf("bad: %v", nodes)
}
if nodes[0].ID != node2.ID ||
nodes[0].Node != "baz" ||
nodes[0].Datacenter != "dc1" ||
nodes[0].Address != "127.0.0.2" ||
len(nodes[0].Meta) != 1 ||
nodes[0].Meta["testMeta"] != "testing123" ||
len(nodes[0].TaggedAddresses) != 1 ||
nodes[0].TaggedAddresses["hello"] != "1.2.3.4" {
t.Fatalf("bad: %v", nodes[0])
}
if nodes[1].ID != node1.ID ||
nodes[1].Node != "foo" ||
nodes[1].Datacenter != "dc1" ||
nodes[1].Address != "127.0.0.1" ||
len(nodes[1].TaggedAddresses) != 0 {
t.Fatalf("bad: %v", nodes[1])
}
_, fooSrv, err := fsm2.state.NodeServices(nil, "foo", nil)
if err != nil {
t.Fatalf("err: %s", err)
}
if len(fooSrv.Services) != 2 {
t.Fatalf("Bad: %v", fooSrv)
}
if !lib.StrContains(fooSrv.Services["db"].Tags, "primary") {
t.Fatalf("Bad: %v", fooSrv)
}
if fooSrv.Services["db"].Port != 5000 {
t.Fatalf("Bad: %v", fooSrv)
}
connectSrv := fooSrv.Services["web"]
if !reflect.DeepEqual(connectSrv.Connect, connectConf) {
t.Fatalf("got: %v, want: %v", connectSrv.Connect, connectConf)
}
_, checks, err := fsm2.state.NodeChecks(nil, "foo", nil)
if err != nil {
t.Fatalf("err: %s", err)
}
if len(checks) != 1 {
t.Fatalf("Bad: %v", checks)
}
// Verify key is set
_, d, err := fsm2.state.KVSGet(nil, "/test", nil)
if err != nil {
t.Fatalf("err: %v", err)
}
if string(d.Value) != "foo" {
t.Fatalf("bad: %v", d)
}
// Verify session is restored
idx, s, err := fsm2.state.SessionGet(nil, session.ID, nil)
if err != nil {
t.Fatalf("err: %v", err)
}
if s.Node != "foo" {
t.Fatalf("bad: %v", s)
}
if idx <= 1 {
t.Fatalf("bad index: %d", idx)
}
// Verify ACL Binding Rule is restored
_, bindingRule2, err := fsm2.state.ACLBindingRuleGetByID(nil, bindingRule.ID, nil)
require.NoError(err)
require.Equal(bindingRule, bindingRule2)
// Verify ACL Auth Method is restored
_, method2, err := fsm2.state.ACLAuthMethodGetByName(nil, method.Name, nil)
require.NoError(err)
require.Equal(method, method2)
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
// Verify ACL Token is restored
_, token2, err := fsm2.state.ACLTokenGetByAccessor(nil, token.AccessorID, nil)
require.NoError(err)
{
// time.Time is tricky to compare generically when it takes a ser/deserialization round trip.
require.True(token.CreateTime.Equal(token2.CreateTime))
token2.CreateTime = token.CreateTime
}
require.Equal(token, token2)
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
// Verify the acl-token-bootstrap index was restored
canBootstrap, index, err := fsm2.state.CanBootstrapACLToken()
require.False(canBootstrap)
require.True(index > 0)
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
// Verify ACL Role is restored
_, role2, err := fsm2.state.ACLRoleGetByID(nil, role.ID, nil)
require.NoError(err)
require.Equal(role, role2)
New ACLs (#4791) This PR is almost a complete rewrite of the ACL system within Consul. It brings the features more in line with other HashiCorp products. Obviously there is quite a bit left to do here but most of it is related docs, testing and finishing the last few commands in the CLI. I will update the PR description and check off the todos as I finish them over the next few days/week. Description At a high level this PR is mainly to split ACL tokens from Policies and to split the concepts of Authorization from Identities. A lot of this PR is mostly just to support CRUD operations on ACLTokens and ACLPolicies. These in and of themselves are not particularly interesting. The bigger conceptual changes are in how tokens get resolved, how backwards compatibility is handled and the separation of policy from identity which could lead the way to allowing for alternative identity providers. On the surface and with a new cluster the ACL system will look very similar to that of Nomads. Both have tokens and policies. Both have local tokens. The ACL management APIs for both are very similar. I even ripped off Nomad's ACL bootstrap resetting procedure. There are a few key differences though. Nomad requires token and policy replication where Consul only requires policy replication with token replication being opt-in. In Consul local tokens only work with token replication being enabled though. All policies in Nomad are globally applicable. In Consul all policies are stored and replicated globally but can be scoped to a subset of the datacenters. This allows for more granular access management. Unlike Nomad, Consul has legacy baggage in the form of the original ACL system. The ramifications of this are: A server running the new system must still support other clients using the legacy system. A client running the new system must be able to use the legacy RPCs when the servers in its datacenter are running the legacy system. The primary ACL DC's servers running in legacy mode needs to be a gate that keeps everything else in the entire multi-DC cluster running in legacy mode. So not only does this PR implement the new ACL system but has a legacy mode built in for when the cluster isn't ready for new ACLs. Also detecting that new ACLs can be used is automatic and requires no configuration on the part of administrators. This process is detailed more in the "Transitioning from Legacy to New ACL Mode" section below.
2018-10-19 16:04:07 +00:00
// Verify ACL Policy is restored
_, policy2, err := fsm2.state.ACLPolicyGetByID(nil, structs.ACLPolicyGlobalManagementID, nil)
require.NoError(err)
require.Equal(policy, policy2)
// Verify tombstones are restored
func() {
snap := fsm2.state.Snapshot()
defer snap.Close()
stones, err := snap.Tombstones()
if err != nil {
t.Fatalf("err: %s", err)
}
stone := stones.Next().(*state.Tombstone)
if stone == nil {
t.Fatalf("missing tombstone")
}
if stone.Key != "/remove" || stone.Index != 12 {
t.Fatalf("bad: %v", stone)
}
if stones.Next() != nil {
t.Fatalf("unexpected extra tombstones")
}
}()
// Verify coordinates are restored
_, coords, err := fsm2.state.Coordinates(nil)
if err != nil {
t.Fatalf("err: %s", err)
}
if !reflect.DeepEqual(coords, updates) {
t.Fatalf("bad: %#v", coords)
}
// Verify queries are restored.
_, queries, err := fsm2.state.PreparedQueryList(nil)
if err != nil {
t.Fatalf("err: %s", err)
}
if len(queries) != 1 {
t.Fatalf("bad: %#v", queries)
}
if !reflect.DeepEqual(queries[0], &query) {
t.Fatalf("bad: %#v", queries[0])
}
// Verify autopilot config is restored.
_, restoredConf, err := fsm2.state.AutopilotConfig()
if err != nil {
t.Fatalf("err: %s", err)
}
if !reflect.DeepEqual(restoredConf, autopilotConf) {
t.Fatalf("bad: %#v, %#v", restoredConf, autopilotConf)
}
// Verify intentions are restored.
_, ixns, err := fsm2.state.Intentions(nil)
require.NoError(err)
assert.Len(ixns, 1)
assert.Equal(ixn, ixns[0])
// Verify CA roots are restored.
_, roots, err = fsm2.state.CARoots(nil)
require.NoError(err)
assert.Len(roots, 2)
// Verify provider state is restored.
_, state, err := fsm2.state.CAProviderState("asdf")
require.NoError(err)
assert.Equal("foo", state.PrivateKey)
assert.Equal("bar", state.RootCert)
// Verify CA configuration is restored.
_, caConf, err := fsm2.state.CAConfig(nil)
require.NoError(err)
assert.Equal(caConfig, caConf)
// Verify config entries are restored
_, serviceConfEntry, err := fsm2.state.ConfigEntry(nil, structs.ServiceDefaults, "foo", structs.DefaultEnterpriseMeta())
require.NoError(err)
assert.Equal(serviceConfig, serviceConfEntry)
_, proxyConfEntry, err := fsm2.state.ConfigEntry(nil, structs.ProxyDefaults, "global", structs.DefaultEnterpriseMeta())
require.NoError(err)
assert.Equal(proxyConfig, proxyConfEntry)
_, ingressRestored, err := fsm2.state.ConfigEntry(nil, structs.IngressGateway, "ingress", structs.DefaultEnterpriseMeta())
require.NoError(err)
assert.Equal(ingress, ingressRestored)
_, restoredGatewayServices, err := fsm2.state.GatewayServices(nil, "ingress", structs.DefaultEnterpriseMeta())
require.NoError(err)
require.Equal(gatewayServices, restoredGatewayServices)
newChunkState, err := fsm2.chunker.CurrentState()
require.NoError(err)
assert.Equal(newChunkState, chunkState)
wan federation via mesh gateways (#6884) This is like a Möbius strip of code due to the fact that low-level components (serf/memberlist) are connected to high-level components (the catalog and mesh-gateways) in a twisty maze of references which make it hard to dive into. With that in mind here's a high level summary of what you'll find in the patch: There are several distinct chunks of code that are affected: * new flags and config options for the server * retry join WAN is slightly different * retry join code is shared to discover primary mesh gateways from secondary datacenters * because retry join logic runs in the *agent* and the results of that operation for primary mesh gateways are needed in the *server* there are some methods like `RefreshPrimaryGatewayFallbackAddresses` that must occur at multiple layers of abstraction just to pass the data down to the right layer. * new cache type `FederationStateListMeshGatewaysName` for use in `proxycfg/xds` layers * the function signature for RPC dialing picked up a new required field (the node name of the destination) * several new RPCs for manipulating a FederationState object: `FederationState:{Apply,Get,List,ListMeshGateways}` * 3 read-only internal APIs for debugging use to invoke those RPCs from curl * raft and fsm changes to persist these FederationStates * replication for FederationStates as they are canonically stored in the Primary and replicated to the Secondaries. * a special derivative of anti-entropy that runs in secondaries to snapshot their local mesh gateway `CheckServiceNodes` and sync them into their upstream FederationState in the primary (this works in conjunction with the replication to distribute addresses for all mesh gateways in all DCs to all other DCs) * a "gateway locator" convenience object to make use of this data to choose the addresses of gateways to use for any given RPC or gossip operation to a remote DC. This gets data from the "retry join" logic in the agent and also directly calls into the FSM. * RPC (`:8300`) on the server sniffs the first byte of a new connection to determine if it's actually doing native TLS. If so it checks the ALPN header for protocol determination (just like how the existing system uses the type-byte marker). * 2 new kinds of protocols are exclusively decoded via this native TLS mechanism: one for ferrying "packet" operations (udp-like) from the gossip layer and one for "stream" operations (tcp-like). The packet operations re-use sockets (using length-prefixing) to cut down on TLS re-negotiation overhead. * the server instances specially wrap the `memberlist.NetTransport` when running with gateway federation enabled (in a `wanfed.Transport`). The general gist is that if it tries to dial a node in the SAME datacenter (deduced by looking at the suffix of the node name) there is no change. If dialing a DIFFERENT datacenter it is wrapped up in a TLS+ALPN blob and sent through some mesh gateways to eventually end up in a server's :8300 port. * a new flag when launching a mesh gateway via `consul connect envoy` to indicate that the servers are to be exposed. This sets a special service meta when registering the gateway into the catalog. * `proxycfg/xds` notice this metadata blob to activate additional watches for the FederationState objects as well as the location of all of the consul servers in that datacenter. * `xds:` if the extra metadata is in place additional clusters are defined in a DC to bulk sink all traffic to another DC's gateways. For the current datacenter we listen on a wildcard name (`server.<dc>.consul`) that load balances all servers as well as one mini-cluster per node (`<node>.server.<dc>.consul`) * the `consul tls cert create` command got a new flag (`-node`) to help create an additional SAN in certs that can be used with this flavor of federation.
2020-03-09 20:59:02 +00:00
// Verify federation states are restored.
_, fedStateLoaded1, err := fsm2.state.FederationStateGet(nil, "dc1")
require.NoError(err)
assert.Equal(fedState1, fedStateLoaded1)
_, fedStateLoaded2, err := fsm2.state.FederationStateGet(nil, "dc2")
require.NoError(err)
assert.Equal(fedState2, fedStateLoaded2)
// Snapshot
snap, err = fsm2.Snapshot()
if err != nil {
t.Fatalf("err: %v", err)
}
defer snap.Release()
// Persist
buf = bytes.NewBuffer(nil)
sink = &MockSink{buf, false}
if err := snap.Persist(sink); err != nil {
t.Fatalf("err: %v", err)
}
// Try to restore on the old FSM and make sure it abandons the old state
// store.
abandonCh := fsm.state.AbandonCh()
if err := fsm.Restore(sink); err != nil {
t.Fatalf("err: %v", err)
}
select {
case <-abandonCh:
default:
t.Fatalf("bad")
}
}
func TestFSM_BadRestore_OSS(t *testing.T) {
t.Parallel()
// Create an FSM with some state.
logger := testutil.Logger(t)
fsm, err := New(nil, logger)
if err != nil {
t.Fatalf("err: %v", err)
}
fsm.state.EnsureNode(1, &structs.Node{Node: "foo", Address: "127.0.0.1"})
abandonCh := fsm.state.AbandonCh()
// Do a bad restore.
buf := bytes.NewBuffer([]byte("bad snapshot"))
sink := &MockSink{buf, false}
if err := fsm.Restore(sink); err == nil {
t.Fatalf("err: %v", err)
}
// Verify the contents didn't get corrupted.
_, nodes, err := fsm.state.Nodes(nil)
if err != nil {
t.Fatalf("err: %s", err)
}
if len(nodes) != 1 {
t.Fatalf("bad: %v", nodes)
}
if nodes[0].Node != "foo" ||
nodes[0].Address != "127.0.0.1" ||
len(nodes[0].TaggedAddresses) != 0 {
t.Fatalf("bad: %v", nodes[0])
}
// Verify the old state store didn't get abandoned.
select {
case <-abandonCh:
t.Fatalf("bad")
default:
}
}
func TestFSM_BadSnapshot_NilCAConfig(t *testing.T) {
t.Parallel()
require := require.New(t)
// Create an FSM with no config entry.
logger := testutil.Logger(t)
fsm, err := New(nil, logger)
if err != nil {
t.Fatalf("err: %v", err)
}
// Snapshot
snap, err := fsm.Snapshot()
if err != nil {
t.Fatalf("err: %v", err)
}
defer snap.Release()
// Persist
buf := bytes.NewBuffer(nil)
sink := &MockSink{buf, false}
if err := snap.Persist(sink); err != nil {
t.Fatalf("err: %v", err)
}
// Try to restore on a new FSM
fsm2, err := New(nil, logger)
if err != nil {
t.Fatalf("err: %v", err)
}
// Do a restore
if err := fsm2.Restore(sink); err != nil {
t.Fatalf("err: %v", err)
}
// Make sure there's no entry in the CA config table.
state := fsm2.State()
idx, config, err := state.CAConfig(nil)
require.NoError(err)
require.Equal(uint64(0), idx)
if config != nil {
t.Fatalf("config should be nil")
}
}