Forward autopilot state reqs, avoid self-dialing (#15493)

Make sure that autopilot is disabled when we step down from active node state.  Forward autopilot state requests to the active node.  Avoid self-dialing due to stale advertisement.

changelog/15493.txt

@@ -0,0 +1,3 @@
+```release-note:bug
+storage/raft: Forward autopilot state requests on perf standbys to active node.
+```

go.mod

@@ -54,6 +54,7 @@ require (
 	github.com/google/go-cmp v0.5.6
 	github.com/google/go-github v17.0.0+incompatible
 	github.com/google/go-metrics-stackdriver v0.2.0
+	github.com/google/tink/go v1.4.0
 	github.com/hashicorp/cap v0.1.1
 	github.com/hashicorp/consul-template v0.29.0
 	github.com/hashicorp/consul/api v1.12.0
@@ -274,7 +275,6 @@ require (
 	github.com/google/go-querystring v1.1.0 // indirect
 	github.com/google/gofuzz v1.1.0 // indirect
 	github.com/google/shlex v0.0.0-20191202100458-e7afc7fbc510 // indirect
-	github.com/google/tink/go v1.4.0 // indirect
 	github.com/google/uuid v1.3.0 // indirect
 	github.com/googleapis/gax-go/v2 v2.0.5 // indirect
 	github.com/googleapis/gnostic v0.5.5 // indirect

go.sum

@@ -885,7 +885,6 @@ github.com/hashicorp/go-secure-stdlib/mlock v0.1.2 h1:p4AKXPPS24tO8Wc8i1gLvSKdmk
 github.com/hashicorp/go-secure-stdlib/mlock v0.1.2/go.mod h1:zq93CJChV6L9QTfGKtfBxKqD7BqqXx5O04A/ns2p5+I=
 github.com/hashicorp/go-secure-stdlib/parseutil v0.1.1/go.mod h1:QmrqtbKuxxSWTN3ETMPuB+VtEiBJ/A9XhoYGv8E1uD8=
 github.com/hashicorp/go-secure-stdlib/parseutil v0.1.2/go.mod h1:QmrqtbKuxxSWTN3ETMPuB+VtEiBJ/A9XhoYGv8E1uD8=
-github.com/hashicorp/go-secure-stdlib/parseutil v0.1.4/go.mod h1:QmrqtbKuxxSWTN3ETMPuB+VtEiBJ/A9XhoYGv8E1uD8=
 github.com/hashicorp/go-secure-stdlib/parseutil v0.1.5 h1:MBgwAFPUbfuI0+tmDU/aeM1MARvdbqWmiieXIalKqDE=
 github.com/hashicorp/go-secure-stdlib/parseutil v0.1.5/go.mod h1:QmrqtbKuxxSWTN3ETMPuB+VtEiBJ/A9XhoYGv8E1uD8=
 github.com/hashicorp/go-secure-stdlib/password v0.1.1 h1:6JzmBqXprakgFEHwBgdchsjaA9x3GyjdI568bXKxa60=

physical/raft/raft_autopilot.go

@@ -499,6 +499,7 @@ func (b *RaftBackend) StopAutopilot() {
 		return
 	}
 	b.autopilot.Stop()
+	b.autopilot = nil
 	b.followerHeartbeatTicker.Stop()
 }
 

vault/ha.go

@@ -185,6 +185,15 @@ func (c *Core) Leader() (isLeader bool, leaderAddr, clusterAddr string, err erro
 		oldAdv = true
 	}
 
+	// At the top of this function we return early when we're the active node.
+	// If we're not the active node, and there's a stale advertisement pointing
+	// to ourself, there's no point in paying any attention to it.  And by
+	// disregarding it, we can avoid a panic in raft tests using the Inmem network
+	// layer when we try to connect back to ourself.
+	if adv.ClusterAddr == c.ClusterAddr() && adv.RedirectAddr == c.redirectAddr {
+		return false, "", "", nil
+	}
+
 	if !oldAdv {
 		c.logger.Debug("parsing information for new active node", "active_cluster_addr", adv.ClusterAddr, "active_redirect_addr", adv.RedirectAddr)
 

vault/logical_system_raft.go

@@ -152,8 +152,9 @@ func (b *SystemBackend) raftStoragePaths() []*framework.Path {
 			Pattern: "storage/raft/autopilot/state",
 			Operations: map[logical.Operation]framework.OperationHandler{
 				logical.ReadOperation: &framework.PathOperation{
-					Callback: b.verifyDROperationTokenOnSecondary(b.handleStorageRaftAutopilotState(), false),
+					Callback:                  b.verifyDROperationTokenOnSecondary(b.handleStorageRaftAutopilotState(), false),
-					Summary:  "Returns the state of the raft cluster under integrated storage as seen by autopilot.",
+					Summary:                   "Returns the state of the raft cluster under integrated storage as seen by autopilot.",
+					ForwardPerformanceStandby: true,
 				},
 			},