open-consul/agent/xds/routes.go

645 lines
20 KiB
Go
Raw Normal View History

package xds
import (
"errors"
"fmt"
"net"
"strings"
2020-09-12 00:34:03 +00:00
"time"
envoy_route_v3 "github.com/envoyproxy/go-control-plane/envoy/config/route/v3"
envoy_matcher_v3 "github.com/envoyproxy/go-control-plane/envoy/type/matcher/v3"
"github.com/golang/protobuf/proto"
"github.com/golang/protobuf/ptypes"
2020-08-28 20:27:40 +00:00
"github.com/hashicorp/consul/agent/connect"
"github.com/hashicorp/consul/agent/proxycfg"
"github.com/hashicorp/consul/agent/structs"
)
// routesFromSnapshot returns the xDS API representation of the "routes" in the
// snapshot.
Support Incremental xDS mode (#9855) This adds support for the Incremental xDS protocol when using xDS v3. This is best reviewed commit-by-commit and will not be squashed when merged. Union of all commit messages follows to give an overarching summary: xds: exclusively support incremental xDS when using xDS v3 Attempts to use SoTW via v3 will fail, much like attempts to use incremental via v2 will fail. Work around a strange older envoy behavior involving empty CDS responses over incremental xDS. xds: various cleanups and refactors that don't strictly concern the addition of incremental xDS support Dissolve the connectionInfo struct in favor of per-connection ResourceGenerators instead. Do a better job of ensuring the xds code uses a well configured logger that accurately describes the connected client. xds: pull out checkStreamACLs method in advance of a later commit xds: rewrite SoTW xDS protocol tests to use protobufs rather than hand-rolled json strings In the test we very lightly reuse some of the more boring protobuf construction helper code that is also technically under test. The important thing of the protocol tests is testing the protocol. The actual inputs and outputs are largely already handled by the xds golden output tests now so these protocol tests don't have to do double-duty. This also updates the SoTW protocol test to exclusively use xDS v2 which is the only variant of SoTW that will be supported in Consul 1.10. xds: default xds.Server.AuthCheckFrequency at use-time instead of construction-time
2021-04-29 18:54:05 +00:00
func (s *ResourceGenerator) routesFromSnapshot(cfgSnap *proxycfg.ConfigSnapshot) ([]proto.Message, error) {
if cfgSnap == nil {
return nil, errors.New("nil config given")
}
switch cfgSnap.Kind {
case structs.ServiceKindConnectProxy:
Support Incremental xDS mode (#9855) This adds support for the Incremental xDS protocol when using xDS v3. This is best reviewed commit-by-commit and will not be squashed when merged. Union of all commit messages follows to give an overarching summary: xds: exclusively support incremental xDS when using xDS v3 Attempts to use SoTW via v3 will fail, much like attempts to use incremental via v2 will fail. Work around a strange older envoy behavior involving empty CDS responses over incremental xDS. xds: various cleanups and refactors that don't strictly concern the addition of incremental xDS support Dissolve the connectionInfo struct in favor of per-connection ResourceGenerators instead. Do a better job of ensuring the xds code uses a well configured logger that accurately describes the connected client. xds: pull out checkStreamACLs method in advance of a later commit xds: rewrite SoTW xDS protocol tests to use protobufs rather than hand-rolled json strings In the test we very lightly reuse some of the more boring protobuf construction helper code that is also technically under test. The important thing of the protocol tests is testing the protocol. The actual inputs and outputs are largely already handled by the xds golden output tests now so these protocol tests don't have to do double-duty. This also updates the SoTW protocol test to exclusively use xDS v2 which is the only variant of SoTW that will be supported in Consul 1.10. xds: default xds.Server.AuthCheckFrequency at use-time instead of construction-time
2021-04-29 18:54:05 +00:00
return s.routesForConnectProxy(cfgSnap.ConnectProxy.DiscoveryChain)
case structs.ServiceKindIngressGateway:
Support Incremental xDS mode (#9855) This adds support for the Incremental xDS protocol when using xDS v3. This is best reviewed commit-by-commit and will not be squashed when merged. Union of all commit messages follows to give an overarching summary: xds: exclusively support incremental xDS when using xDS v3 Attempts to use SoTW via v3 will fail, much like attempts to use incremental via v2 will fail. Work around a strange older envoy behavior involving empty CDS responses over incremental xDS. xds: various cleanups and refactors that don't strictly concern the addition of incremental xDS support Dissolve the connectionInfo struct in favor of per-connection ResourceGenerators instead. Do a better job of ensuring the xds code uses a well configured logger that accurately describes the connected client. xds: pull out checkStreamACLs method in advance of a later commit xds: rewrite SoTW xDS protocol tests to use protobufs rather than hand-rolled json strings In the test we very lightly reuse some of the more boring protobuf construction helper code that is also technically under test. The important thing of the protocol tests is testing the protocol. The actual inputs and outputs are largely already handled by the xds golden output tests now so these protocol tests don't have to do double-duty. This also updates the SoTW protocol test to exclusively use xDS v2 which is the only variant of SoTW that will be supported in Consul 1.10. xds: default xds.Server.AuthCheckFrequency at use-time instead of construction-time
2021-04-29 18:54:05 +00:00
return s.routesForIngressGateway(cfgSnap.IngressGateway.Upstreams, cfgSnap.IngressGateway.DiscoveryChain)
2020-08-28 20:27:40 +00:00
case structs.ServiceKindTerminatingGateway:
Support Incremental xDS mode (#9855) This adds support for the Incremental xDS protocol when using xDS v3. This is best reviewed commit-by-commit and will not be squashed when merged. Union of all commit messages follows to give an overarching summary: xds: exclusively support incremental xDS when using xDS v3 Attempts to use SoTW via v3 will fail, much like attempts to use incremental via v2 will fail. Work around a strange older envoy behavior involving empty CDS responses over incremental xDS. xds: various cleanups and refactors that don't strictly concern the addition of incremental xDS support Dissolve the connectionInfo struct in favor of per-connection ResourceGenerators instead. Do a better job of ensuring the xds code uses a well configured logger that accurately describes the connected client. xds: pull out checkStreamACLs method in advance of a later commit xds: rewrite SoTW xDS protocol tests to use protobufs rather than hand-rolled json strings In the test we very lightly reuse some of the more boring protobuf construction helper code that is also technically under test. The important thing of the protocol tests is testing the protocol. The actual inputs and outputs are largely already handled by the xds golden output tests now so these protocol tests don't have to do double-duty. This also updates the SoTW protocol test to exclusively use xDS v2 which is the only variant of SoTW that will be supported in Consul 1.10. xds: default xds.Server.AuthCheckFrequency at use-time instead of construction-time
2021-04-29 18:54:05 +00:00
return s.routesFromSnapshotTerminatingGateway(cfgSnap)
case structs.ServiceKindMeshGateway:
return nil, nil // mesh gateways will never have routes
default:
return nil, fmt.Errorf("Invalid service kind: %v", cfgSnap.Kind)
}
}
2020-09-11 16:49:26 +00:00
// routesFromSnapshotConnectProxy returns the xDS API representation of the
// "routes" in the snapshot.
Support Incremental xDS mode (#9855) This adds support for the Incremental xDS protocol when using xDS v3. This is best reviewed commit-by-commit and will not be squashed when merged. Union of all commit messages follows to give an overarching summary: xds: exclusively support incremental xDS when using xDS v3 Attempts to use SoTW via v3 will fail, much like attempts to use incremental via v2 will fail. Work around a strange older envoy behavior involving empty CDS responses over incremental xDS. xds: various cleanups and refactors that don't strictly concern the addition of incremental xDS support Dissolve the connectionInfo struct in favor of per-connection ResourceGenerators instead. Do a better job of ensuring the xds code uses a well configured logger that accurately describes the connected client. xds: pull out checkStreamACLs method in advance of a later commit xds: rewrite SoTW xDS protocol tests to use protobufs rather than hand-rolled json strings In the test we very lightly reuse some of the more boring protobuf construction helper code that is also technically under test. The important thing of the protocol tests is testing the protocol. The actual inputs and outputs are largely already handled by the xds golden output tests now so these protocol tests don't have to do double-duty. This also updates the SoTW protocol test to exclusively use xDS v2 which is the only variant of SoTW that will be supported in Consul 1.10. xds: default xds.Server.AuthCheckFrequency at use-time instead of construction-time
2021-04-29 18:54:05 +00:00
func (s *ResourceGenerator) routesForConnectProxy(chains map[string]*structs.CompiledDiscoveryChain) ([]proto.Message, error) {
2020-09-11 16:49:26 +00:00
var resources []proto.Message
2021-03-17 19:40:49 +00:00
for id, chain := range chains {
if chain.IsDefault() {
continue
2020-09-11 16:49:26 +00:00
}
Support Incremental xDS mode (#9855) This adds support for the Incremental xDS protocol when using xDS v3. This is best reviewed commit-by-commit and will not be squashed when merged. Union of all commit messages follows to give an overarching summary: xds: exclusively support incremental xDS when using xDS v3 Attempts to use SoTW via v3 will fail, much like attempts to use incremental via v2 will fail. Work around a strange older envoy behavior involving empty CDS responses over incremental xDS. xds: various cleanups and refactors that don't strictly concern the addition of incremental xDS support Dissolve the connectionInfo struct in favor of per-connection ResourceGenerators instead. Do a better job of ensuring the xds code uses a well configured logger that accurately describes the connected client. xds: pull out checkStreamACLs method in advance of a later commit xds: rewrite SoTW xDS protocol tests to use protobufs rather than hand-rolled json strings In the test we very lightly reuse some of the more boring protobuf construction helper code that is also technically under test. The important thing of the protocol tests is testing the protocol. The actual inputs and outputs are largely already handled by the xds golden output tests now so these protocol tests don't have to do double-duty. This also updates the SoTW protocol test to exclusively use xDS v2 which is the only variant of SoTW that will be supported in Consul 1.10. xds: default xds.Server.AuthCheckFrequency at use-time instead of construction-time
2021-04-29 18:54:05 +00:00
virtualHost, err := makeUpstreamRouteForDiscoveryChain(id, chain, []string{"*"})
2021-03-17 19:40:49 +00:00
if err != nil {
return nil, err
}
2020-09-11 16:49:26 +00:00
2021-03-17 19:40:49 +00:00
route := &envoy_route_v3.RouteConfiguration{
Name: id,
VirtualHosts: []*envoy_route_v3.VirtualHost{virtualHost},
// ValidateClusters defaults to true when defined statically and false
// when done via RDS. Re-set the sane value of true to prevent
// null-routing traffic.
ValidateClusters: makeBoolValue(true),
2020-09-11 16:49:26 +00:00
}
2021-03-17 19:40:49 +00:00
resources = append(resources, route)
2020-09-11 16:49:26 +00:00
}
// TODO(rb): make sure we don't generate an empty result
return resources, nil
}
2020-08-28 20:27:40 +00:00
// routesFromSnapshotTerminatingGateway returns the xDS API representation of the "routes" in the snapshot.
// For any HTTP service we will return a default route.
Support Incremental xDS mode (#9855) This adds support for the Incremental xDS protocol when using xDS v3. This is best reviewed commit-by-commit and will not be squashed when merged. Union of all commit messages follows to give an overarching summary: xds: exclusively support incremental xDS when using xDS v3 Attempts to use SoTW via v3 will fail, much like attempts to use incremental via v2 will fail. Work around a strange older envoy behavior involving empty CDS responses over incremental xDS. xds: various cleanups and refactors that don't strictly concern the addition of incremental xDS support Dissolve the connectionInfo struct in favor of per-connection ResourceGenerators instead. Do a better job of ensuring the xds code uses a well configured logger that accurately describes the connected client. xds: pull out checkStreamACLs method in advance of a later commit xds: rewrite SoTW xDS protocol tests to use protobufs rather than hand-rolled json strings In the test we very lightly reuse some of the more boring protobuf construction helper code that is also technically under test. The important thing of the protocol tests is testing the protocol. The actual inputs and outputs are largely already handled by the xds golden output tests now so these protocol tests don't have to do double-duty. This also updates the SoTW protocol test to exclusively use xDS v2 which is the only variant of SoTW that will be supported in Consul 1.10. xds: default xds.Server.AuthCheckFrequency at use-time instead of construction-time
2021-04-29 18:54:05 +00:00
func (s *ResourceGenerator) routesFromSnapshotTerminatingGateway(cfgSnap *proxycfg.ConfigSnapshot) ([]proto.Message, error) {
if cfgSnap == nil {
return nil, errors.New("nil config given")
}
var resources []proto.Message
for _, svc := range cfgSnap.TerminatingGateway.ValidServices() {
2020-08-28 20:27:40 +00:00
clusterName := connect.ServiceSNI(svc.Name, "", svc.NamespaceOrDefault(), cfgSnap.Datacenter, cfgSnap.Roots.TrustDomain)
resolver, hasResolver := cfgSnap.TerminatingGateway.ServiceResolvers[svc]
svcConfig := cfgSnap.TerminatingGateway.ServiceConfigs[svc]
cfg, err := ParseProxyConfig(svcConfig.ProxyConfig)
2020-09-03 16:21:20 +00:00
if err != nil {
return nil, fmt.Errorf("failed to parse upstream config: %v", err)
}
if !structs.IsProtocolHTTPLike(cfg.Protocol) {
2020-08-28 20:27:40 +00:00
// Routes can only be defined for HTTP services
continue
}
if !hasResolver {
// Use a zero value resolver with no timeout and no subsets
resolver = &structs.ServiceResolverConfigEntry{}
}
2020-09-11 15:21:43 +00:00
var lb *structs.LoadBalancer
if resolver.LoadBalancer != nil {
2020-09-11 15:21:43 +00:00
lb = resolver.LoadBalancer
}
route, err := makeNamedDefaultRouteWithLB(clusterName, lb, true)
2020-08-28 20:27:40 +00:00
if err != nil {
Support Incremental xDS mode (#9855) This adds support for the Incremental xDS protocol when using xDS v3. This is best reviewed commit-by-commit and will not be squashed when merged. Union of all commit messages follows to give an overarching summary: xds: exclusively support incremental xDS when using xDS v3 Attempts to use SoTW via v3 will fail, much like attempts to use incremental via v2 will fail. Work around a strange older envoy behavior involving empty CDS responses over incremental xDS. xds: various cleanups and refactors that don't strictly concern the addition of incremental xDS support Dissolve the connectionInfo struct in favor of per-connection ResourceGenerators instead. Do a better job of ensuring the xds code uses a well configured logger that accurately describes the connected client. xds: pull out checkStreamACLs method in advance of a later commit xds: rewrite SoTW xDS protocol tests to use protobufs rather than hand-rolled json strings In the test we very lightly reuse some of the more boring protobuf construction helper code that is also technically under test. The important thing of the protocol tests is testing the protocol. The actual inputs and outputs are largely already handled by the xds golden output tests now so these protocol tests don't have to do double-duty. This also updates the SoTW protocol test to exclusively use xDS v2 which is the only variant of SoTW that will be supported in Consul 1.10. xds: default xds.Server.AuthCheckFrequency at use-time instead of construction-time
2021-04-29 18:54:05 +00:00
s.Logger.Error("failed to make route", "cluster", clusterName, "error", err)
2020-08-28 20:27:40 +00:00
continue
}
resources = append(resources, route)
// If there is a service-resolver for this service then also setup routes for each subset
for name := range resolver.Subsets {
clusterName = connect.ServiceSNI(svc.Name, name, svc.NamespaceOrDefault(), cfgSnap.Datacenter, cfgSnap.Roots.TrustDomain)
route, err := makeNamedDefaultRouteWithLB(clusterName, lb, true)
2020-08-28 20:27:40 +00:00
if err != nil {
Support Incremental xDS mode (#9855) This adds support for the Incremental xDS protocol when using xDS v3. This is best reviewed commit-by-commit and will not be squashed when merged. Union of all commit messages follows to give an overarching summary: xds: exclusively support incremental xDS when using xDS v3 Attempts to use SoTW via v3 will fail, much like attempts to use incremental via v2 will fail. Work around a strange older envoy behavior involving empty CDS responses over incremental xDS. xds: various cleanups and refactors that don't strictly concern the addition of incremental xDS support Dissolve the connectionInfo struct in favor of per-connection ResourceGenerators instead. Do a better job of ensuring the xds code uses a well configured logger that accurately describes the connected client. xds: pull out checkStreamACLs method in advance of a later commit xds: rewrite SoTW xDS protocol tests to use protobufs rather than hand-rolled json strings In the test we very lightly reuse some of the more boring protobuf construction helper code that is also technically under test. The important thing of the protocol tests is testing the protocol. The actual inputs and outputs are largely already handled by the xds golden output tests now so these protocol tests don't have to do double-duty. This also updates the SoTW protocol test to exclusively use xDS v2 which is the only variant of SoTW that will be supported in Consul 1.10. xds: default xds.Server.AuthCheckFrequency at use-time instead of construction-time
2021-04-29 18:54:05 +00:00
s.Logger.Error("failed to make route", "cluster", clusterName, "error", err)
2020-08-28 20:27:40 +00:00
continue
}
resources = append(resources, route)
}
}
return resources, nil
}
func makeNamedDefaultRouteWithLB(clusterName string, lb *structs.LoadBalancer, autoHostRewrite bool) (*envoy_route_v3.RouteConfiguration, error) {
2020-08-28 20:27:40 +00:00
action := makeRouteActionFromName(clusterName)
if err := injectLBToRouteAction(lb, action.Route); err != nil {
2020-08-28 20:27:40 +00:00
return nil, fmt.Errorf("failed to apply load balancer configuration to route action: %v", err)
}
// Configure Envoy to rewrite Host header
if autoHostRewrite {
action.Route.HostRewriteSpecifier = &envoy_route_v3.RouteAction_AutoHostRewrite{
AutoHostRewrite: makeBoolValue(true),
}
}
return &envoy_route_v3.RouteConfiguration{
2020-08-28 20:27:40 +00:00
Name: clusterName,
VirtualHosts: []*envoy_route_v3.VirtualHost{
2020-08-28 20:27:40 +00:00
{
Name: clusterName,
Domains: []string{"*"},
Routes: []*envoy_route_v3.Route{
2020-08-28 20:27:40 +00:00
{
Match: makeDefaultRouteMatch(),
Action: action,
},
},
},
},
// ValidateClusters defaults to true when defined statically and false
// when done via RDS. Re-set the sane value of true to prevent
// null-routing traffic.
ValidateClusters: makeBoolValue(true),
}, nil
}
// routesForIngressGateway returns the xDS API representation of the
// "routes" in the snapshot.
Support Incremental xDS mode (#9855) This adds support for the Incremental xDS protocol when using xDS v3. This is best reviewed commit-by-commit and will not be squashed when merged. Union of all commit messages follows to give an overarching summary: xds: exclusively support incremental xDS when using xDS v3 Attempts to use SoTW via v3 will fail, much like attempts to use incremental via v2 will fail. Work around a strange older envoy behavior involving empty CDS responses over incremental xDS. xds: various cleanups and refactors that don't strictly concern the addition of incremental xDS support Dissolve the connectionInfo struct in favor of per-connection ResourceGenerators instead. Do a better job of ensuring the xds code uses a well configured logger that accurately describes the connected client. xds: pull out checkStreamACLs method in advance of a later commit xds: rewrite SoTW xDS protocol tests to use protobufs rather than hand-rolled json strings In the test we very lightly reuse some of the more boring protobuf construction helper code that is also technically under test. The important thing of the protocol tests is testing the protocol. The actual inputs and outputs are largely already handled by the xds golden output tests now so these protocol tests don't have to do double-duty. This also updates the SoTW protocol test to exclusively use xDS v2 which is the only variant of SoTW that will be supported in Consul 1.10. xds: default xds.Server.AuthCheckFrequency at use-time instead of construction-time
2021-04-29 18:54:05 +00:00
func (s *ResourceGenerator) routesForIngressGateway(
2020-08-28 20:27:40 +00:00
upstreams map[proxycfg.IngressListenerKey]structs.Upstreams,
chains map[string]*structs.CompiledDiscoveryChain,
) ([]proto.Message, error) {
var result []proto.Message
2020-08-28 20:27:40 +00:00
for listenerKey, upstreams := range upstreams {
// Do not create any route configuration for TCP listeners
if listenerKey.Protocol == "tcp" {
continue
}
upstreamRoute := &envoy_route_v3.RouteConfiguration{
Name: listenerKey.RouteName(),
// ValidateClusters defaults to true when defined statically and false
// when done via RDS. Re-set the sane value of true to prevent
// null-routing traffic.
ValidateClusters: makeBoolValue(true),
}
for _, u := range upstreams {
upstreamID := u.Identifier()
2020-08-28 20:27:40 +00:00
chain := chains[upstreamID]
if chain == nil {
continue
}
domains := generateUpstreamIngressDomains(listenerKey, u)
Support Incremental xDS mode (#9855) This adds support for the Incremental xDS protocol when using xDS v3. This is best reviewed commit-by-commit and will not be squashed when merged. Union of all commit messages follows to give an overarching summary: xds: exclusively support incremental xDS when using xDS v3 Attempts to use SoTW via v3 will fail, much like attempts to use incremental via v2 will fail. Work around a strange older envoy behavior involving empty CDS responses over incremental xDS. xds: various cleanups and refactors that don't strictly concern the addition of incremental xDS support Dissolve the connectionInfo struct in favor of per-connection ResourceGenerators instead. Do a better job of ensuring the xds code uses a well configured logger that accurately describes the connected client. xds: pull out checkStreamACLs method in advance of a later commit xds: rewrite SoTW xDS protocol tests to use protobufs rather than hand-rolled json strings In the test we very lightly reuse some of the more boring protobuf construction helper code that is also technically under test. The important thing of the protocol tests is testing the protocol. The actual inputs and outputs are largely already handled by the xds golden output tests now so these protocol tests don't have to do double-duty. This also updates the SoTW protocol test to exclusively use xDS v2 which is the only variant of SoTW that will be supported in Consul 1.10. xds: default xds.Server.AuthCheckFrequency at use-time instead of construction-time
2021-04-29 18:54:05 +00:00
virtualHost, err := makeUpstreamRouteForDiscoveryChain(upstreamID, chain, domains)
if err != nil {
return nil, err
}
upstreamRoute.VirtualHosts = append(upstreamRoute.VirtualHosts, virtualHost)
}
result = append(result, upstreamRoute)
}
return result, nil
}
func generateUpstreamIngressDomains(listenerKey proxycfg.IngressListenerKey, u structs.Upstream) []string {
var domains []string
domainsSet := make(map[string]bool)
namespace := u.GetEnterpriseMeta().NamespaceOrDefault()
switch {
case len(u.IngressHosts) > 0:
// If a user has specified hosts, do not add the default
// "<service-name>.ingress.*" prefixes
domains = u.IngressHosts
case namespace != structs.IntentionDefaultNamespace:
domains = []string{fmt.Sprintf("%s.ingress.%s.*", u.DestinationName, namespace)}
default:
domains = []string{fmt.Sprintf("%s.ingress.*", u.DestinationName)}
}
for _, h := range domains {
domainsSet[h] = true
}
// Host headers may contain port numbers in them, so we need to make sure
// we match on the host with and without the port number. Well-known
// ports like HTTP/HTTPS are stripped from Host headers, but other ports
// will be in the header.
for _, h := range domains {
_, _, err := net.SplitHostPort(h)
// Error message from Go's net/ipsock.go
// We check to see if a port is not missing, and ignore the
// error from SplitHostPort otherwise, since we have previously
// validated the Host values and should trust the user's input.
if err == nil || !strings.Contains(err.Error(), "missing port in address") {
continue
}
domainWithPort := fmt.Sprintf("%s:%d", h, listenerKey.Port)
// Do not add a duplicate domain if a hostname with port is already in the
// set
if !domainsSet[domainWithPort] {
domains = append(domains, domainWithPort)
}
}
return domains
}
func makeUpstreamRouteForDiscoveryChain(
routeName string,
chain *structs.CompiledDiscoveryChain,
serviceDomains []string,
) (*envoy_route_v3.VirtualHost, error) {
var routes []*envoy_route_v3.Route
startNode := chain.Nodes[chain.StartNode]
if startNode == nil {
connect: use stronger validation that ingress gateways have compatible protocols defined for their upstreams (#8470) Fixes #8466 Since Consul 1.8.0 there was a bug in how ingress gateway protocol compatibility was enforced. At the point in time that an ingress-gateway config entry was modified the discovery chain for each upstream was checked to ensure the ingress gateway protocol matched. Unfortunately future modifications of other config entries were not validated against existing ingress-gateway definitions, such as: 1. create tcp ingress-gateway pointing to 'api' (ok) 2. create service-defaults for 'api' setting protocol=http (worked, but not ok) 3. create service-splitter or service-router for 'api' (worked, but caused an agent panic) If you were to do these in a different order, it would fail without a crash: 1. create service-defaults for 'api' setting protocol=http (ok) 2. create service-splitter or service-router for 'api' (ok) 3. create tcp ingress-gateway pointing to 'api' (fail with message about protocol mismatch) This PR introduces the missing validation. The two new behaviors are: 1. create tcp ingress-gateway pointing to 'api' (ok) 2. (NEW) create service-defaults for 'api' setting protocol=http ("ok" for back compat) 3. (NEW) create service-splitter or service-router for 'api' (fail with message about protocol mismatch) In consideration for any existing users that may be inadvertently be falling into item (2) above, that is now officiall a valid configuration to be in. For anyone falling into item (3) above while you cannot use the API to manufacture that scenario anymore, anyone that has old (now bad) data will still be able to have the agent use them just enough to generate a new agent/proxycfg error message rather than a panic. Unfortunately we just don't have enough information to properly fix the config entries.
2020-08-12 16:19:20 +00:00
return nil, fmt.Errorf("missing first node in compiled discovery chain for: %s", chain.ServiceName)
}
switch startNode.Type {
case structs.DiscoveryGraphNodeTypeRouter:
routes = make([]*envoy_route_v3.Route, 0, len(startNode.Routes))
for _, discoveryRoute := range startNode.Routes {
Support Incremental xDS mode (#9855) This adds support for the Incremental xDS protocol when using xDS v3. This is best reviewed commit-by-commit and will not be squashed when merged. Union of all commit messages follows to give an overarching summary: xds: exclusively support incremental xDS when using xDS v3 Attempts to use SoTW via v3 will fail, much like attempts to use incremental via v2 will fail. Work around a strange older envoy behavior involving empty CDS responses over incremental xDS. xds: various cleanups and refactors that don't strictly concern the addition of incremental xDS support Dissolve the connectionInfo struct in favor of per-connection ResourceGenerators instead. Do a better job of ensuring the xds code uses a well configured logger that accurately describes the connected client. xds: pull out checkStreamACLs method in advance of a later commit xds: rewrite SoTW xDS protocol tests to use protobufs rather than hand-rolled json strings In the test we very lightly reuse some of the more boring protobuf construction helper code that is also technically under test. The important thing of the protocol tests is testing the protocol. The actual inputs and outputs are largely already handled by the xds golden output tests now so these protocol tests don't have to do double-duty. This also updates the SoTW protocol test to exclusively use xDS v2 which is the only variant of SoTW that will be supported in Consul 1.10. xds: default xds.Server.AuthCheckFrequency at use-time instead of construction-time
2021-04-29 18:54:05 +00:00
routeMatch := makeRouteMatchForDiscoveryRoute(discoveryRoute)
var (
routeAction *envoy_route_v3.Route_Route
err error
)
nextNode := chain.Nodes[discoveryRoute.NextNode]
2020-09-11 16:49:26 +00:00
var lb *structs.LoadBalancer
if nextNode.LoadBalancer != nil {
2020-09-11 15:21:43 +00:00
lb = nextNode.LoadBalancer
}
switch nextNode.Type {
case structs.DiscoveryGraphNodeTypeSplitter:
routeAction, err = makeRouteActionForSplitter(nextNode.Splits, chain)
if err != nil {
return nil, err
}
if err := injectLBToRouteAction(lb, routeAction.Route); err != nil {
2020-08-28 20:27:40 +00:00
return nil, fmt.Errorf("failed to apply load balancer configuration to route action: %v", err)
}
case structs.DiscoveryGraphNodeTypeResolver:
2020-08-28 20:27:40 +00:00
routeAction = makeRouteActionForChainCluster(nextNode.Resolver.Target, chain)
if err := injectLBToRouteAction(lb, routeAction.Route); err != nil {
2020-08-28 20:27:40 +00:00
return nil, fmt.Errorf("failed to apply load balancer configuration to route action: %v", err)
}
default:
return nil, fmt.Errorf("unexpected graph node after route %q", nextNode.Type)
}
// TODO(rb): Better help handle the envoy case where you need (prefix=/foo/,rewrite=/) and (exact=/foo,rewrite=/) to do a full rewrite
destination := discoveryRoute.Definition.Destination
if destination != nil {
if destination.PrefixRewrite != "" {
routeAction.Route.PrefixRewrite = destination.PrefixRewrite
}
if destination.RequestTimeout > 0 {
routeAction.Route.Timeout = ptypes.DurationProto(destination.RequestTimeout)
}
if destination.HasRetryFeatures() {
retryPolicy := &envoy_route_v3.RetryPolicy{}
if destination.NumRetries > 0 {
retryPolicy.NumRetries = makeUint32Value(int(destination.NumRetries))
}
// The RetryOn magic values come from: https://www.envoyproxy.io/docs/envoy/v1.10.0/configuration/http_filters/router_filter#config-http-filters-router-x-envoy-retry-on
if destination.RetryOnConnectFailure {
retryPolicy.RetryOn = "connect-failure"
}
if len(destination.RetryOnStatusCodes) > 0 {
if retryPolicy.RetryOn != "" {
retryPolicy.RetryOn = retryPolicy.RetryOn + ",retriable-status-codes"
} else {
retryPolicy.RetryOn = "retriable-status-codes"
}
retryPolicy.RetriableStatusCodes = destination.RetryOnStatusCodes
}
routeAction.Route.RetryPolicy = retryPolicy
}
}
routes = append(routes, &envoy_route_v3.Route{
Match: routeMatch,
Action: routeAction,
})
}
case structs.DiscoveryGraphNodeTypeSplitter:
routeAction, err := makeRouteActionForSplitter(startNode.Splits, chain)
if err != nil {
return nil, err
}
2020-09-11 16:49:26 +00:00
var lb *structs.LoadBalancer
if startNode.LoadBalancer != nil {
2020-09-11 15:21:43 +00:00
lb = startNode.LoadBalancer
}
if err := injectLBToRouteAction(lb, routeAction.Route); err != nil {
2020-08-28 20:27:40 +00:00
return nil, fmt.Errorf("failed to apply load balancer configuration to route action: %v", err)
}
defaultRoute := &envoy_route_v3.Route{
Match: makeDefaultRouteMatch(),
Action: routeAction,
}
routes = []*envoy_route_v3.Route{defaultRoute}
case structs.DiscoveryGraphNodeTypeResolver:
2020-08-28 20:27:40 +00:00
routeAction := makeRouteActionForChainCluster(startNode.Resolver.Target, chain)
2020-09-11 16:49:26 +00:00
var lb *structs.LoadBalancer
if startNode.LoadBalancer != nil {
2020-09-11 15:21:43 +00:00
lb = startNode.LoadBalancer
}
if err := injectLBToRouteAction(lb, routeAction.Route); err != nil {
2020-08-28 20:27:40 +00:00
return nil, fmt.Errorf("failed to apply load balancer configuration to route action: %v", err)
}
defaultRoute := &envoy_route_v3.Route{
Match: makeDefaultRouteMatch(),
Action: routeAction,
}
routes = []*envoy_route_v3.Route{defaultRoute}
default:
connect: use stronger validation that ingress gateways have compatible protocols defined for their upstreams (#8470) Fixes #8466 Since Consul 1.8.0 there was a bug in how ingress gateway protocol compatibility was enforced. At the point in time that an ingress-gateway config entry was modified the discovery chain for each upstream was checked to ensure the ingress gateway protocol matched. Unfortunately future modifications of other config entries were not validated against existing ingress-gateway definitions, such as: 1. create tcp ingress-gateway pointing to 'api' (ok) 2. create service-defaults for 'api' setting protocol=http (worked, but not ok) 3. create service-splitter or service-router for 'api' (worked, but caused an agent panic) If you were to do these in a different order, it would fail without a crash: 1. create service-defaults for 'api' setting protocol=http (ok) 2. create service-splitter or service-router for 'api' (ok) 3. create tcp ingress-gateway pointing to 'api' (fail with message about protocol mismatch) This PR introduces the missing validation. The two new behaviors are: 1. create tcp ingress-gateway pointing to 'api' (ok) 2. (NEW) create service-defaults for 'api' setting protocol=http ("ok" for back compat) 3. (NEW) create service-splitter or service-router for 'api' (fail with message about protocol mismatch) In consideration for any existing users that may be inadvertently be falling into item (2) above, that is now officiall a valid configuration to be in. For anyone falling into item (3) above while you cannot use the API to manufacture that scenario anymore, anyone that has old (now bad) data will still be able to have the agent use them just enough to generate a new agent/proxycfg error message rather than a panic. Unfortunately we just don't have enough information to properly fix the config entries.
2020-08-12 16:19:20 +00:00
return nil, fmt.Errorf("unknown first node in discovery chain of type: %s", startNode.Type)
}
host := &envoy_route_v3.VirtualHost{
Name: routeName,
Domains: serviceDomains,
Routes: routes,
}
return host, nil
}
Support Incremental xDS mode (#9855) This adds support for the Incremental xDS protocol when using xDS v3. This is best reviewed commit-by-commit and will not be squashed when merged. Union of all commit messages follows to give an overarching summary: xds: exclusively support incremental xDS when using xDS v3 Attempts to use SoTW via v3 will fail, much like attempts to use incremental via v2 will fail. Work around a strange older envoy behavior involving empty CDS responses over incremental xDS. xds: various cleanups and refactors that don't strictly concern the addition of incremental xDS support Dissolve the connectionInfo struct in favor of per-connection ResourceGenerators instead. Do a better job of ensuring the xds code uses a well configured logger that accurately describes the connected client. xds: pull out checkStreamACLs method in advance of a later commit xds: rewrite SoTW xDS protocol tests to use protobufs rather than hand-rolled json strings In the test we very lightly reuse some of the more boring protobuf construction helper code that is also technically under test. The important thing of the protocol tests is testing the protocol. The actual inputs and outputs are largely already handled by the xds golden output tests now so these protocol tests don't have to do double-duty. This also updates the SoTW protocol test to exclusively use xDS v2 which is the only variant of SoTW that will be supported in Consul 1.10. xds: default xds.Server.AuthCheckFrequency at use-time instead of construction-time
2021-04-29 18:54:05 +00:00
func makeRouteMatchForDiscoveryRoute(discoveryRoute *structs.DiscoveryRoute) *envoy_route_v3.RouteMatch {
match := discoveryRoute.Definition.Match
if match == nil || match.IsEmpty() {
return makeDefaultRouteMatch()
}
em := &envoy_route_v3.RouteMatch{}
switch {
case match.HTTP.PathExact != "":
em.PathSpecifier = &envoy_route_v3.RouteMatch_Path{
Path: match.HTTP.PathExact,
}
case match.HTTP.PathPrefix != "":
em.PathSpecifier = &envoy_route_v3.RouteMatch_Prefix{
Prefix: match.HTTP.PathPrefix,
}
case match.HTTP.PathRegex != "":
em.PathSpecifier = &envoy_route_v3.RouteMatch_SafeRegex{
SafeRegex: makeEnvoyRegexMatch(match.HTTP.PathRegex),
}
default:
em.PathSpecifier = &envoy_route_v3.RouteMatch_Prefix{
Prefix: "/",
}
}
if len(match.HTTP.Header) > 0 {
em.Headers = make([]*envoy_route_v3.HeaderMatcher, 0, len(match.HTTP.Header))
for _, hdr := range match.HTTP.Header {
eh := &envoy_route_v3.HeaderMatcher{
Name: hdr.Name,
}
switch {
case hdr.Exact != "":
eh.HeaderMatchSpecifier = &envoy_route_v3.HeaderMatcher_ExactMatch{
ExactMatch: hdr.Exact,
}
case hdr.Regex != "":
eh.HeaderMatchSpecifier = &envoy_route_v3.HeaderMatcher_SafeRegexMatch{
SafeRegexMatch: makeEnvoyRegexMatch(hdr.Regex),
}
case hdr.Prefix != "":
eh.HeaderMatchSpecifier = &envoy_route_v3.HeaderMatcher_PrefixMatch{
PrefixMatch: hdr.Prefix,
}
case hdr.Suffix != "":
eh.HeaderMatchSpecifier = &envoy_route_v3.HeaderMatcher_SuffixMatch{
SuffixMatch: hdr.Suffix,
}
case hdr.Present:
eh.HeaderMatchSpecifier = &envoy_route_v3.HeaderMatcher_PresentMatch{
PresentMatch: true,
}
default:
continue // skip this impossible situation
}
if hdr.Invert {
eh.InvertMatch = true
}
em.Headers = append(em.Headers, eh)
}
}
if len(match.HTTP.Methods) > 0 {
methodHeaderRegex := strings.Join(match.HTTP.Methods, "|")
eh := &envoy_route_v3.HeaderMatcher{
Name: ":method",
HeaderMatchSpecifier: &envoy_route_v3.HeaderMatcher_SafeRegexMatch{
SafeRegexMatch: makeEnvoyRegexMatch(methodHeaderRegex),
},
}
em.Headers = append(em.Headers, eh)
}
if len(match.HTTP.QueryParam) > 0 {
em.QueryParameters = make([]*envoy_route_v3.QueryParameterMatcher, 0, len(match.HTTP.QueryParam))
for _, qm := range match.HTTP.QueryParam {
eq := &envoy_route_v3.QueryParameterMatcher{
Name: qm.Name,
}
switch {
case qm.Exact != "":
eq.QueryParameterMatchSpecifier = &envoy_route_v3.QueryParameterMatcher_StringMatch{
StringMatch: &envoy_matcher_v3.StringMatcher{
MatchPattern: &envoy_matcher_v3.StringMatcher_Exact{
Exact: qm.Exact,
},
},
}
case qm.Regex != "":
eq.QueryParameterMatchSpecifier = &envoy_route_v3.QueryParameterMatcher_StringMatch{
StringMatch: &envoy_matcher_v3.StringMatcher{
MatchPattern: &envoy_matcher_v3.StringMatcher_SafeRegex{
SafeRegex: makeEnvoyRegexMatch(qm.Regex),
},
},
}
case qm.Present:
eq.QueryParameterMatchSpecifier = &envoy_route_v3.QueryParameterMatcher_PresentMatch{
PresentMatch: true,
}
default:
continue // skip this impossible situation
}
em.QueryParameters = append(em.QueryParameters, eq)
}
}
return em
}
func makeDefaultRouteMatch() *envoy_route_v3.RouteMatch {
return &envoy_route_v3.RouteMatch{
PathSpecifier: &envoy_route_v3.RouteMatch_Prefix{
Prefix: "/",
},
// TODO(banks) Envoy supports matching only valid GRPC
// requests which might be nice to add here for gRPC services
// but it's not supported in our current envoy SDK version
// although docs say it was supported by 1.8.0. Going to defer
// that until we've updated the deps.
}
}
func makeRouteActionForChainCluster(targetID string, chain *structs.CompiledDiscoveryChain) *envoy_route_v3.Route_Route {
target := chain.Targets[targetID]
2020-08-28 20:27:40 +00:00
return makeRouteActionFromName(CustomizeClusterName(target.Name, chain))
}
func makeRouteActionFromName(clusterName string) *envoy_route_v3.Route_Route {
return &envoy_route_v3.Route_Route{
Route: &envoy_route_v3.RouteAction{
ClusterSpecifier: &envoy_route_v3.RouteAction_Cluster{
Cluster: clusterName,
},
},
}
}
func makeRouteActionForSplitter(splits []*structs.DiscoverySplit, chain *structs.CompiledDiscoveryChain) (*envoy_route_v3.Route_Route, error) {
clusters := make([]*envoy_route_v3.WeightedCluster_ClusterWeight, 0, len(splits))
for _, split := range splits {
nextNode := chain.Nodes[split.NextNode]
if nextNode.Type != structs.DiscoveryGraphNodeTypeResolver {
return nil, fmt.Errorf("unexpected splitter destination node type: %s", nextNode.Type)
}
targetID := nextNode.Resolver.Target
target := chain.Targets[targetID]
connect: reconcile how upstream configuration works with discovery chains (#6225) * connect: reconcile how upstream configuration works with discovery chains The following upstream config fields for connect sidecars sanely integrate into discovery chain resolution: - Destination Namespace/Datacenter: Compilation occurs locally but using different default values for namespaces and datacenters. The xDS clusters that are created are named as they normally would be. - Mesh Gateway Mode (single upstream): If set this value overrides any value computed for any resolver for the entire discovery chain. The xDS clusters that are created may be named differently (see below). - Mesh Gateway Mode (whole sidecar): If set this value overrides any value computed for any resolver for the entire discovery chain. If this is specifically overridden for a single upstream this value is ignored in that case. The xDS clusters that are created may be named differently (see below). - Protocol (in opaque config): If set this value overrides the value computed when evaluating the entire discovery chain. If the normal chain would be TCP or if this override is set to TCP then the result is that we explicitly disable L7 Routing and Splitting. The xDS clusters that are created may be named differently (see below). - Connect Timeout (in opaque config): If set this value overrides the value for any resolver in the entire discovery chain. The xDS clusters that are created may be named differently (see below). If any of the above overrides affect the actual result of compiling the discovery chain (i.e. "tcp" becomes "grpc" instead of being a no-op override to "tcp") then the relevant parameters are hashed and provided to the xDS layer as a prefix for use in naming the Clusters. This is to ensure that if one Upstream discovery chain has no overrides and tangentially needs a cluster named "api.default.XXX", and another Upstream does have overrides for "api.default.XXX" that they won't cross-pollinate against the operator's wishes. Fixes #6159
2019-08-02 03:03:34 +00:00
clusterName := CustomizeClusterName(target.Name, chain)
// The smallest representable weight is 1/10000 or .01% but envoy
// deals with integers so scale everything up by 100x.
cw := &envoy_route_v3.WeightedCluster_ClusterWeight{
Weight: makeUint32Value(int(split.Weight * 100)),
Name: clusterName,
}
clusters = append(clusters, cw)
}
return &envoy_route_v3.Route_Route{
Route: &envoy_route_v3.RouteAction{
ClusterSpecifier: &envoy_route_v3.RouteAction_WeightedClusters{
WeightedClusters: &envoy_route_v3.WeightedCluster{
Clusters: clusters,
TotalWeight: makeUint32Value(10000), // scaled up 100%
},
},
},
}, nil
}
func injectLBToRouteAction(lb *structs.LoadBalancer, action *envoy_route_v3.RouteAction) error {
2020-09-11 15:21:43 +00:00
if lb == nil || !lb.IsHashBased() {
return nil
}
result := make([]*envoy_route_v3.RouteAction_HashPolicy, 0, len(lb.HashPolicies))
2020-09-11 15:21:43 +00:00
for _, policy := range lb.HashPolicies {
if policy.SourceIP {
result = append(result, &envoy_route_v3.RouteAction_HashPolicy{
PolicySpecifier: &envoy_route_v3.RouteAction_HashPolicy_ConnectionProperties_{
ConnectionProperties: &envoy_route_v3.RouteAction_HashPolicy_ConnectionProperties{
SourceIp: true,
},
},
Terminal: policy.Terminal,
})
continue
}
switch policy.Field {
case structs.HashPolicyHeader:
result = append(result, &envoy_route_v3.RouteAction_HashPolicy{
PolicySpecifier: &envoy_route_v3.RouteAction_HashPolicy_Header_{
Header: &envoy_route_v3.RouteAction_HashPolicy_Header{
HeaderName: policy.FieldValue,
},
},
Terminal: policy.Terminal,
})
case structs.HashPolicyCookie:
cookie := envoy_route_v3.RouteAction_HashPolicy_Cookie{
Name: policy.FieldValue,
}
if policy.CookieConfig != nil {
cookie.Path = policy.CookieConfig.Path
2020-09-12 00:34:03 +00:00
if policy.CookieConfig.TTL != 0*time.Second {
cookie.Ttl = ptypes.DurationProto(policy.CookieConfig.TTL)
}
// Envoy will generate a session cookie if the ttl is present and zero.
if policy.CookieConfig.Session {
cookie.Ttl = ptypes.DurationProto(0 * time.Second)
}
}
result = append(result, &envoy_route_v3.RouteAction_HashPolicy{
PolicySpecifier: &envoy_route_v3.RouteAction_HashPolicy_Cookie_{
Cookie: &cookie,
},
Terminal: policy.Terminal,
})
case structs.HashPolicyQueryParam:
result = append(result, &envoy_route_v3.RouteAction_HashPolicy{
PolicySpecifier: &envoy_route_v3.RouteAction_HashPolicy_QueryParameter_{
QueryParameter: &envoy_route_v3.RouteAction_HashPolicy_QueryParameter{
Name: policy.FieldValue,
},
},
Terminal: policy.Terminal,
})
default:
return fmt.Errorf("unsupported load balancer hash policy field: %v", policy.Field)
}
}
action.HashPolicy = result
return nil
}