open-consul/agent/proxycfg/state.go
R.B. Boyer b089f93292
proxycfg: ensure that an irrecoverable error in proxycfg closes the xds session and triggers a replacement proxycfg watcher (#16497)
Receiving an "acl not found" error from an RPC in the agent cache and the
streaming/event components will cause any request loops to cease under the
assumption that they will never work again if the token was destroyed. This
prevents log spam (#14144, #9738).

Unfortunately due to things like:

- authz requests going to stale servers that may not have witnessed the token
  creation yet

- authz requests in a secondary datacenter happening before the tokens get
  replicated to that datacenter

- authz requests from a primary TO a secondary datacenter happening before the
  tokens get replicated to that datacenter

The caller will get an "acl not found" *before* the token exists, rather than
just after. The machinery added above in the linked PRs will kick in and
prevent the request loop from looping around again once the tokens actually
exist.

For `consul-dataplane` usages, where xDS is served by the Consul servers
rather than the clients ultimately this is not a problem because in that
scenario the `agent/proxycfg` machinery is on-demand and launched by a new xDS
stream needing data for a specific service in the catalog. If the watching
goroutines are terminated it ripples down and terminates the xDS stream, which
CDP will eventually re-establish and restart everything.

For Consul client usages, the `agent/proxycfg` machinery is ahead-of-time
launched at service registration time (called "local" in some of the proxycfg
machinery) so when the xDS stream comes in the data is already ready to go. If
the watching goroutines terminate it should terminate the xDS stream, but
there's no mechanism to re-spawn the watching goroutines. If the xDS stream
reconnects it will see no `ConfigSnapshot` and will not get one again until
the client agent is restarted, or the service is re-registered with something
changed in it.

This PR fixes a few things in the machinery:

- there was an inadvertent deadlock in fetching snapshot from the proxycfg
  machinery by xDS, such that when the watching goroutine terminated the
  snapshots would never be fetched. This caused some of the xDS machinery to
  get indefinitely paused and not finish the teardown properly.

- Every 30s we now attempt to re-insert all locally registered services into
  the proxycfg machinery.

- When services are re-inserted into the proxycfg machinery we special case
  "dead" ones such that we unilaterally replace them rather that doing that
  conditionally.
2023-03-03 14:27:53 -06:00

542 lines
17 KiB
Go

package proxycfg
import (
"context"
"errors"
"fmt"
"net"
"reflect"
"runtime/debug"
"sync/atomic"
"time"
"github.com/hashicorp/go-hclog"
"golang.org/x/time/rate"
cachetype "github.com/hashicorp/consul/agent/cache-types"
"github.com/hashicorp/consul/agent/structs"
"github.com/hashicorp/consul/logging"
)
const (
coalesceTimeout = 200 * time.Millisecond
rootsWatchID = "roots"
peeringTrustBundlesWatchID = "peering-trust-bundles"
leafWatchID = "leaf"
peerTrustBundleIDPrefix = "peer-trust-bundle:"
intentionsWatchID = "intentions"
serviceListWatchID = "service-list"
peeringServiceListWatchID = "peering-service-list:"
federationStateListGatewaysWatchID = "federation-state-list-mesh-gateways"
consulServerListWatchID = "consul-server-list"
datacentersWatchID = "datacenters"
serviceResolversWatchID = "service-resolvers"
gatewayServicesWatchID = "gateway-services"
gatewayConfigWatchID = "gateway-config"
inlineCertificateConfigWatchID = "inline-certificate-config"
routeConfigWatchID = "route-config"
externalServiceIDPrefix = "external-service:"
serviceLeafIDPrefix = "service-leaf:"
serviceConfigIDPrefix = "service-config:"
serviceResolverIDPrefix = "service-resolver:"
serviceIntentionsIDPrefix = "service-intentions:"
intentionUpstreamsID = "intention-upstreams"
peerServersWatchID = "peer-servers"
peeredUpstreamsID = "peered-upstreams"
intentionUpstreamsDestinationID = "intention-upstreams-destination"
upstreamPeerWatchIDPrefix = "upstream-peer:"
exportedServiceListWatchID = "exported-service-list"
meshConfigEntryID = "mesh"
DestinationConfigEntryID = "destination:"
DestinationGatewayID = "dest-gateway:"
svcChecksWatchIDPrefix = cachetype.ServiceHTTPChecksName + ":"
preparedQueryIDPrefix = string(structs.UpstreamDestTypePreparedQuery) + ":"
defaultPreparedQueryPollInterval = 30 * time.Second
)
type stateConfig struct {
logger hclog.Logger
source *structs.QuerySource
dataSources DataSources
dnsConfig DNSConfig
serverSNIFn ServerSNIFunc
intentionDefaultAllow bool
}
// state holds all the state needed to maintain the config for a registered
// connect-proxy service. When a proxy registration is changed, the entire state
// is discarded and a new one created.
type state struct {
source ProxySource
logger hclog.Logger
serviceInstance serviceInstance
handler kindHandler
// cancel is set by Watch and called by Close to stop the goroutine started
// in Watch.
cancel func()
// failedFlag is (atomically) set to 1 (by Close) when run exits because a data
// source is in an irrecoverable state. It can be read with failed.
failedFlag int32
ch chan UpdateEvent
snapCh chan ConfigSnapshot
reqCh chan chan *ConfigSnapshot
doneCh chan struct{}
rateLimiter *rate.Limiter
}
func (s *state) stoppedRunning() bool {
select {
case <-s.doneCh:
return true
default:
return false
}
}
// failed returns whether run exited because a data source is in an
// irrecoverable state.
func (s *state) failed() bool {
return atomic.LoadInt32(&s.failedFlag) == 1
}
type DNSConfig struct {
Domain string
AltDomain string
}
type ServerSNIFunc func(dc, nodeName string) string
type serviceInstance struct {
kind structs.ServiceKind
service string
proxyID ProxyID
address string
port int
meta map[string]string
taggedAddresses map[string]structs.ServiceAddress
proxyCfg structs.ConnectProxyConfig
token string
}
func copyProxyConfig(ns *structs.NodeService) (structs.ConnectProxyConfig, error) {
if ns == nil {
return structs.ConnectProxyConfig{}, nil
}
proxyCfg := *(&ns.Proxy).DeepCopy()
// we can safely modify these since we just copied them
for idx := range proxyCfg.Upstreams {
us := &proxyCfg.Upstreams[idx]
if us.DestinationType != structs.UpstreamDestTypePreparedQuery {
// default the upstreams target namespace and partition to those of the proxy
// doing this here prevents needing much more complex logic a bunch of other
// places and makes tracking these upstreams simpler as we can dedup them
// with the maps tracking upstream ids being watched.
if us.DestinationPartition == "" {
proxyCfg.Upstreams[idx].DestinationPartition = ns.EnterpriseMeta.PartitionOrDefault()
}
if us.DestinationNamespace == "" {
proxyCfg.Upstreams[idx].DestinationNamespace = ns.EnterpriseMeta.NamespaceOrDefault()
}
// If PeerName is not empty, the DestinationPartition refers
// to the local Partition in which the Peer exists and the
// DestinationNamespace refers to the Namespace residing in
// the remote peer
if us.DestinationPeer == "" {
proxyCfg.Upstreams[idx].DestinationPeer = ns.PeerName
}
}
}
return proxyCfg, nil
}
// newState populates the state struct by copying relevant fields from the
// NodeService and Token. We copy so that we can use them in a separate
// goroutine later without reasoning about races with the NodeService passed
// (especially for embedded fields like maps and slices).
//
// The returned state needs its required dependencies to be set before Watch
// can be called.
func newState(id ProxyID, ns *structs.NodeService, source ProxySource, token string, config stateConfig, rateLimiter *rate.Limiter) (*state, error) {
// 10 is fairly arbitrary here but allow for the 3 mandatory and a
// reasonable number of upstream watches to all deliver their initial
// messages in parallel without blocking the cache.Notify loops. It's not a
// huge deal if we do for a short period so we don't need to be more
// conservative to handle larger numbers of upstreams correctly but gives
// some head room for normal operation to be non-blocking in most typical
// cases.
ch := make(chan UpdateEvent, 10)
s, err := newServiceInstanceFromNodeService(id, ns, token)
if err != nil {
return nil, err
}
handler, err := newKindHandler(config, s, ch)
if err != nil {
return nil, err
}
return &state{
source: source,
logger: config.logger.With("proxy", s.proxyID, "kind", s.kind),
serviceInstance: s,
handler: handler,
ch: ch,
snapCh: make(chan ConfigSnapshot, 1),
reqCh: make(chan chan *ConfigSnapshot, 1),
doneCh: make(chan struct{}),
rateLimiter: rateLimiter,
}, nil
}
func newKindHandler(config stateConfig, s serviceInstance, ch chan UpdateEvent) (kindHandler, error) {
var handler kindHandler
h := handlerState{stateConfig: config, serviceInstance: s, ch: ch}
switch s.kind {
case structs.ServiceKindConnectProxy:
handler = &handlerConnectProxy{handlerState: h}
case structs.ServiceKindTerminatingGateway:
h.stateConfig.logger = config.logger.Named(logging.TerminatingGateway)
handler = &handlerTerminatingGateway{handlerState: h}
case structs.ServiceKindMeshGateway:
h.stateConfig.logger = config.logger.Named(logging.MeshGateway)
handler = &handlerMeshGateway{handlerState: h}
case structs.ServiceKindIngressGateway:
handler = &handlerIngressGateway{handlerState: h}
case structs.ServiceKindAPIGateway:
handler = &handlerAPIGateway{handlerState: h}
default:
return nil, errors.New("not a connect-proxy, terminating-gateway, mesh-gateway, or ingress-gateway")
}
return handler, nil
}
func newServiceInstanceFromNodeService(id ProxyID, ns *structs.NodeService, token string) (serviceInstance, error) {
proxyCfg, err := copyProxyConfig(ns)
if err != nil {
return serviceInstance{}, err
}
taggedAddresses := make(map[string]structs.ServiceAddress)
for k, v := range ns.TaggedAddresses {
taggedAddresses[k] = v
}
meta := make(map[string]string)
for k, v := range ns.Meta {
meta[k] = v
}
return serviceInstance{
kind: ns.Kind,
service: ns.Service,
proxyID: id,
address: ns.Address,
port: ns.Port,
meta: meta,
taggedAddresses: taggedAddresses,
proxyCfg: proxyCfg,
token: token,
}, nil
}
type kindHandler interface {
initialize(ctx context.Context) (ConfigSnapshot, error)
handleUpdate(ctx context.Context, u UpdateEvent, snap *ConfigSnapshot) error
}
// Watch initialized watches on all necessary cache data for the current proxy
// registration state and returns a chan to observe updates to the
// ConfigSnapshot that contains all necessary config state. The chan is closed
// when the state is Closed.
func (s *state) Watch() (<-chan ConfigSnapshot, error) {
var ctx context.Context
ctx, s.cancel = context.WithCancel(context.Background())
snap, err := s.handler.initialize(ctx)
if err != nil {
s.cancel()
return nil, err
}
go s.run(ctx, &snap)
return s.snapCh, nil
}
// Close discards the state and stops any long-running watches.
func (s *state) Close(failed bool) error {
if s.stoppedRunning() {
return nil
}
if s.cancel != nil {
s.cancel()
}
if failed {
atomic.StoreInt32(&s.failedFlag, 1)
}
return nil
}
type handlerState struct {
stateConfig // TODO: un-embed
serviceInstance // TODO: un-embed
ch chan UpdateEvent
}
func newConfigSnapshotFromServiceInstance(s serviceInstance, config stateConfig) ConfigSnapshot {
// TODO: use serviceInstance type in ConfigSnapshot
return ConfigSnapshot{
Kind: s.kind,
Service: s.service,
ProxyID: s.proxyID,
Address: s.address,
Port: s.port,
ServiceMeta: s.meta,
TaggedAddresses: s.taggedAddresses,
Proxy: s.proxyCfg,
Datacenter: config.source.Datacenter,
Locality: GatewayKey{Datacenter: config.source.Datacenter, Partition: s.proxyID.PartitionOrDefault()},
ServerSNIFn: config.serverSNIFn,
IntentionDefaultAllow: config.intentionDefaultAllow,
}
}
func (s *state) run(ctx context.Context, snap *ConfigSnapshot) {
// Add a recover here so than any panics do not make their way up
// into the server / agent.
defer func() {
if r := recover(); r != nil {
s.logger.Error("unexpected panic while running proxycfg",
"node", s.serviceInstance.proxyID.NodeName,
"service", s.serviceInstance.proxyID.ServiceID,
"message", r,
"stacktrace", string(debug.Stack()))
}
}()
s.unsafeRun(ctx, snap)
}
func (s *state) unsafeRun(ctx context.Context, snap *ConfigSnapshot) {
// Closing the done channel signals that this entire state is no longer
// going to be updated.
defer close(s.doneCh)
// Close the channel we return from Watch when we stop so consumers can stop
// watching and clean up their goroutines. It's important we do this here and
// not in Close since this routine sends on this chan and so might panic if it
// gets closed from another goroutine.
defer close(s.snapCh)
// This turns out to be really fiddly/painful by just using time.Timer.C
// directly in the code below since you can't detect when a timer is stopped
// vs waiting in order to know to reset it. So just use a chan to send
// ourselves messages.
sendCh := make(chan struct{})
var coalesceTimer *time.Timer
scheduleUpdate := func() {
// Wait for MAX(<rate limiter delay>, coalesceTimeout)
delay := s.rateLimiter.Reserve().Delay()
if delay < coalesceTimeout {
delay = coalesceTimeout
}
coalesceTimer = time.AfterFunc(delay, func() {
// This runs in another goroutine so we can't just do the send
// directly here as access to snap is racy. Instead, signal the main
// loop above.
select {
case sendCh <- struct{}{}:
case <-ctx.Done():
}
})
}
for {
select {
case <-ctx.Done():
return
case u := <-s.ch:
s.logger.Trace("Data source returned; handling snapshot update", "correlationID", u.CorrelationID)
if IsTerminalError(u.Err) {
s.logger.Error("Data source in an irrecoverable state; exiting", "error", u.Err, "correlationID", u.CorrelationID)
s.Close(true)
return
}
if err := s.handler.handleUpdate(ctx, u, snap); err != nil {
s.logger.Error("Failed to handle update from watch",
"id", u.CorrelationID, "error", err,
)
continue
}
case <-sendCh:
// Allow the next change to trigger a send
coalesceTimer = nil
// Make a deep copy of snap so we don't mutate any of the embedded structs
// etc on future updates.
snapCopy := snap.Clone()
select {
// Try to send
case s.snapCh <- *snapCopy:
s.logger.Trace("Delivered new snapshot to proxy config watchers")
// Skip rest of loop - there is nothing to send since nothing changed on
// this iteration
continue
// Avoid blocking if a snapshot is already buffered in snapCh as this can result in a deadlock.
// See PR #9689 for more details.
default:
s.logger.Trace("Failed to deliver new snapshot to proxy config watchers")
// Reset the timer to retry later. This is to ensure we attempt to redeliver the updated snapshot shortly.
scheduleUpdate()
// Do not reset coalesceTimer since we just queued a timer-based refresh
continue
}
case replyCh := <-s.reqCh:
s.logger.Trace("A proxy config snapshot was requested")
if !snap.Valid() {
// Not valid yet just respond with nil and move on to next task.
replyCh <- nil
s.logger.Trace("The proxy's config snapshot is not valid yet")
continue
}
// Make a deep copy of snap so we don't mutate any of the embedded structs
// etc on future updates.
replyCh <- snap.Clone()
// Skip rest of loop - there is nothing to send since nothing changed on
// this iteration
continue
}
// Check if snap is complete enough to be a valid config to deliver to a
// proxy yet.
if snap.Valid() {
if coalesceTimer == nil {
// Don't send it right away, set a short timer that will wait for updates
// from any of the other cache values and deliver them all together.
scheduleUpdate()
}
}
}
}
// CurrentSnapshot synchronously returns the current ConfigSnapshot if there is
// one ready. If we don't have one yet because not all necessary parts have been
// returned (i.e. both roots and leaf cert), nil is returned.
func (s *state) CurrentSnapshot() *ConfigSnapshot {
// Make a chan for the response to be sent on
ch := make(chan *ConfigSnapshot, 1)
select {
case <-s.doneCh:
return nil
case s.reqCh <- ch:
}
// Wait for the response
select {
case <-s.doneCh:
return nil
case resp := <-ch:
return resp
}
}
// Changed returns whether or not the passed NodeService has had any of the
// fields we care about for config state watching changed or a different token.
func (s *state) Changed(ns *structs.NodeService, token string) bool {
if ns == nil {
return true
}
proxyCfg, err := copyProxyConfig(ns)
if err != nil {
s.logger.Warn("Failed to parse proxy config and will treat the new service as unchanged")
}
i := s.serviceInstance
return ns.Kind != i.kind ||
i.address != ns.Address ||
i.port != ns.Port ||
!reflect.DeepEqual(i.proxyCfg, proxyCfg) ||
i.token != token
}
// hostnameEndpoints returns all CheckServiceNodes that have hostnames instead of IPs as the address.
// Envoy cannot resolve hostnames provided through EDS, so we exclusively use CDS for these clusters.
// If there is a mix of hostnames and addresses we exclusively use the hostnames, since clusters cannot discover
// services with both EDS and DNS.
func hostnameEndpoints(logger hclog.Logger, localKey GatewayKey, nodes structs.CheckServiceNodes) structs.CheckServiceNodes {
var (
hasIP bool
hasHostname bool
resp structs.CheckServiceNodes
)
for _, n := range nodes {
_, addr, _ := n.BestAddress(!localKey.Matches(n.Node.Datacenter, n.Node.PartitionOrDefault()))
if net.ParseIP(addr) != nil {
hasIP = true
continue
}
hasHostname = true
resp = append(resp, n)
}
if hasHostname && hasIP {
dc := nodes[0].Node.Datacenter
sn := nodes[0].Service.CompoundServiceName()
logger.Warn("service contains instances with mix of hostnames and IP addresses; only hostnames will be passed to Envoy",
"dc", dc, "service", sn.String())
}
return resp
}
type gatewayWatchOpts struct {
internalServiceDump InternalServiceDump
notifyCh chan UpdateEvent
source structs.QuerySource
token string
key GatewayKey
upstreamID UpstreamID
}
func watchMeshGateway(ctx context.Context, opts gatewayWatchOpts) error {
var correlationId string
if opts.upstreamID.Name == "" {
correlationId = fmt.Sprintf("mesh-gateway:%s", opts.key.String())
} else {
correlationId = fmt.Sprintf("mesh-gateway:%s:%s", opts.key.String(), opts.upstreamID.String())
}
return opts.internalServiceDump.Notify(ctx, &structs.ServiceDumpRequest{
Datacenter: opts.key.Datacenter,
QueryOptions: structs.QueryOptions{Token: opts.token},
ServiceKind: structs.ServiceKindMeshGateway,
UseServiceKind: true,
Source: opts.source,
EnterpriseMeta: *structs.DefaultEnterpriseMetaInPartition(opts.key.Partition),
}, correlationId, opts.notifyCh)
}