Skip to content

Commit fc67199

Browse files
rboyerhc-github-team-consul-core
authored andcommitted
health: ensure /v1/health/service/:service endpoint returns the most recent results when a filter is used with streaming (#12640)
The primary bug here is in the streaming subsystem that makes the overall v1/health/service/:service request behave incorrectly when servicing a blocking request with a filter provided. There is a secondary non-streaming bug being fixed here that is much less obvious related to when to update the `reply` variable in a `blockingQuery` evaluation. It is unlikely that it is triggerable in practical environments and I could not actually get the bug to manifest, but I fixed it anyway while investigating the original issue. Simple reproduction (streaming): 1. Register a service with a tag. curl -sL --request PUT 'http://localhost:8500/v1/agent/service/register' \ --header 'Content-Type: application/json' \ --data-raw '{ "ID": "ID1", "Name": "test", "Tags":[ "a" ], "EnableTagOverride": true }' 2. Do an initial filter query that matches on the tag. curl -sLi --get 'http://localhost:8500/v1/health/service/test' --data-urlencode 'filter=a in Service.Tags' 3. Note you get one result. Use the `X-Consul-Index` header to establish a blocking query in another terminal, this should not return yet. curl -sLi --get 'http://localhost:8500/v1/health/service/test?index=$INDEX' --data-urlencode 'filter=a in Service.Tags' 4. Re-register that service with a different tag. curl -sL --request PUT 'http://localhost:8500/v1/agent/service/register' \ --header 'Content-Type: application/json' \ --data-raw '{ "ID": "ID1", "Name": "test", "Tags":[ "b" ], "EnableTagOverride": true }' 5. Your blocking query from (3) should return with a header `X-Consul-Query-Backend: streaming` and empty results if it works correctly `[]`. Attempts to reproduce with non-streaming failed (where you add `&near=_agent` to the read queries and ensure `X-Consul-Query-Backend: blocking-query` shows up in the results).
1 parent 756f5e8 commit fc67199

File tree

6 files changed

+232
-9
lines changed

6 files changed

+232
-9
lines changed

Diff for: .changelog/12640.txt

+3
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,3 @@
1+
```release-note:bug
2+
health: ensure /v1/health/service/:service endpoint returns the most recent results when a filter is used with streaming #12640
3+
```

Diff for: agent/consul/health_endpoint.go

+14-6
Original file line numberDiff line numberDiff line change
@@ -236,30 +236,38 @@ func (h *Health) ServiceNodes(args *structs.ServiceSpecificRequest, reply *struc
236236
&args.QueryOptions,
237237
&reply.QueryMeta,
238238
func(ws memdb.WatchSet, state *state.Store) error {
239+
var thisReply structs.IndexedCheckServiceNodes
240+
239241
index, nodes, err := f(ws, state, args)
240242
if err != nil {
241243
return err
242244
}
243245

244-
reply.Index, reply.Nodes = index, nodes
246+
thisReply.Index, thisReply.Nodes = index, nodes
247+
245248
if len(args.NodeMetaFilters) > 0 {
246-
reply.Nodes = nodeMetaFilter(args.NodeMetaFilters, reply.Nodes)
249+
thisReply.Nodes = nodeMetaFilter(args.NodeMetaFilters, thisReply.Nodes)
247250
}
248251

249-
raw, err := filter.Execute(reply.Nodes)
252+
raw, err := filter.Execute(thisReply.Nodes)
250253
if err != nil {
251254
return err
252255
}
253-
reply.Nodes = raw.(structs.CheckServiceNodes)
256+
thisReply.Nodes = raw.(structs.CheckServiceNodes)
254257

255258
// Note: we filter the results with ACLs *after* applying the user-supplied
256259
// bexpr filter, to ensure QueryMeta.ResultsFilteredByACLs does not include
257260
// results that would be filtered out even if the user did have permission.
258-
if err := h.srv.filterACL(args.Token, reply); err != nil {
261+
if err := h.srv.filterACL(args.Token, &thisReply); err != nil {
259262
return err
260263
}
261264

262-
return h.srv.sortNodesByDistanceFrom(args.Source, reply.Nodes)
265+
if err := h.srv.sortNodesByDistanceFrom(args.Source, thisReply.Nodes); err != nil {
266+
return err
267+
}
268+
269+
*reply = thisReply
270+
return nil
263271
})
264272

265273
// Provide some metrics

Diff for: agent/consul/health_endpoint_test.go

+79
Original file line numberDiff line numberDiff line change
@@ -679,6 +679,85 @@ func TestHealth_ServiceNodes(t *testing.T) {
679679
}
680680
}
681681

682+
func TestHealth_ServiceNodes_BlockingQuery_withFilter(t *testing.T) {
683+
if testing.Short() {
684+
t.Skip("too slow for testing.Short")
685+
}
686+
687+
t.Parallel()
688+
689+
_, s1 := testServer(t)
690+
codec := rpcClient(t, s1)
691+
692+
waitForLeaderEstablishment(t, s1)
693+
694+
register := func(t *testing.T, name, tag string) {
695+
arg := structs.RegisterRequest{
696+
Datacenter: "dc1",
697+
ID: types.NodeID("43d419c0-433b-42c3-bf8a-193eba0b41a3"),
698+
Node: "node1",
699+
Address: "127.0.0.1",
700+
Service: &structs.NodeService{
701+
ID: name,
702+
Service: name,
703+
Tags: []string{tag},
704+
},
705+
}
706+
var out struct{}
707+
require.NoError(t, msgpackrpc.CallWithCodec(codec, "Catalog.Register", &arg, &out))
708+
}
709+
710+
register(t, "web", "foo")
711+
712+
var lastIndex uint64
713+
runStep(t, "read original", func(t *testing.T) {
714+
var out structs.IndexedCheckServiceNodes
715+
req := structs.ServiceSpecificRequest{
716+
Datacenter: "dc1",
717+
ServiceName: "web",
718+
QueryOptions: structs.QueryOptions{
719+
Filter: "foo in Service.Tags",
720+
},
721+
}
722+
require.NoError(t, msgpackrpc.CallWithCodec(codec, "Health.ServiceNodes", &req, &out))
723+
724+
require.Len(t, out.Nodes, 1)
725+
node := out.Nodes[0]
726+
require.Equal(t, "node1", node.Node.Node)
727+
require.Equal(t, "web", node.Service.Service)
728+
require.Equal(t, []string{"foo"}, node.Service.Tags)
729+
730+
require.Equal(t, structs.QueryBackendBlocking, out.Backend)
731+
lastIndex = out.Index
732+
})
733+
734+
runStep(t, "read blocking query result", func(t *testing.T) {
735+
req := structs.ServiceSpecificRequest{
736+
Datacenter: "dc1",
737+
ServiceName: "web",
738+
QueryOptions: structs.QueryOptions{
739+
Filter: "foo in Service.Tags",
740+
},
741+
}
742+
req.MinQueryIndex = lastIndex
743+
744+
var out structs.IndexedCheckServiceNodes
745+
errCh := channelCallRPC(s1, "Health.ServiceNodes", &req, &out, nil)
746+
747+
time.Sleep(200 * time.Millisecond)
748+
749+
// Change the tags
750+
register(t, "web", "bar")
751+
752+
if err := <-errCh; err != nil {
753+
require.NoError(t, err)
754+
}
755+
756+
require.Equal(t, structs.QueryBackendBlocking, out.Backend)
757+
require.Len(t, out.Nodes, 0)
758+
})
759+
}
760+
682761
func TestHealth_ServiceNodes_MultipleServiceTags(t *testing.T) {
683762
if testing.Short() {
684763
t.Skip("too slow for testing.Short")

Diff for: agent/health_endpoint_test.go

+130
Original file line numberDiff line numberDiff line change
@@ -20,6 +20,7 @@ import (
2020
"github.com/hashicorp/consul/api"
2121
"github.com/hashicorp/consul/sdk/testutil/retry"
2222
"github.com/hashicorp/consul/testrpc"
23+
"github.com/hashicorp/consul/types"
2324
)
2425

2526
func TestHealthChecksInState(t *testing.T) {
@@ -902,6 +903,135 @@ use_streaming_backend = true
902903
}
903904
}
904905

906+
func TestHealthServiceNodes_Blocking_withFilter(t *testing.T) {
907+
cases := []struct {
908+
name string
909+
hcl string
910+
queryBackend string
911+
}{
912+
{
913+
name: "no streaming",
914+
queryBackend: "blocking-query",
915+
hcl: `use_streaming_backend = false`,
916+
},
917+
{
918+
name: "streaming",
919+
hcl: `
920+
rpc { enable_streaming = true }
921+
use_streaming_backend = true
922+
`,
923+
queryBackend: "streaming",
924+
},
925+
}
926+
927+
runStep := func(t *testing.T, name string, fn func(t *testing.T)) {
928+
t.Helper()
929+
if !t.Run(name, fn) {
930+
t.FailNow()
931+
}
932+
}
933+
934+
register := func(t *testing.T, a *TestAgent, name, tag string) {
935+
args := &structs.RegisterRequest{
936+
Datacenter: "dc1",
937+
ID: types.NodeID("43d419c0-433b-42c3-bf8a-193eba0b41a3"),
938+
Node: "node1",
939+
Address: "127.0.0.1",
940+
Service: &structs.NodeService{
941+
ID: name,
942+
Service: name,
943+
Tags: []string{tag},
944+
},
945+
}
946+
947+
var out struct{}
948+
require.NoError(t, a.RPC("Catalog.Register", args, &out))
949+
}
950+
951+
for _, tc := range cases {
952+
tc := tc
953+
t.Run(tc.name, func(t *testing.T) {
954+
a := NewTestAgent(t, tc.hcl)
955+
defer a.Shutdown()
956+
testrpc.WaitForTestAgent(t, a.RPC, "dc1")
957+
958+
// Register one with a tag.
959+
register(t, a, "web", "foo")
960+
961+
filterUrlPart := "filter=" + url.QueryEscape("foo in Service.Tags")
962+
963+
// TODO: use other call format
964+
965+
// Initial request with a filter should return one.
966+
var lastIndex uint64
967+
runStep(t, "read original", func(t *testing.T) {
968+
req, err := http.NewRequest("GET", "/v1/health/service/web?dc=dc1&"+filterUrlPart, nil)
969+
require.NoError(t, err)
970+
971+
resp := httptest.NewRecorder()
972+
obj, err := a.srv.HealthServiceNodes(resp, req)
973+
require.NoError(t, err)
974+
975+
nodes := obj.(structs.CheckServiceNodes)
976+
977+
require.Len(t, nodes, 1)
978+
979+
node := nodes[0]
980+
require.Equal(t, "node1", node.Node.Node)
981+
require.Equal(t, "web", node.Service.Service)
982+
require.Equal(t, []string{"foo"}, node.Service.Tags)
983+
984+
require.Equal(t, "blocking-query", resp.Header().Get("X-Consul-Query-Backend"))
985+
986+
idx := getIndex(t, resp)
987+
require.True(t, idx > 0)
988+
989+
lastIndex = idx
990+
})
991+
992+
const timeout = 30 * time.Second
993+
runStep(t, "read blocking query result", func(t *testing.T) {
994+
var (
995+
// out and resp are not safe to read until reading from errCh
996+
out structs.CheckServiceNodes
997+
resp = httptest.NewRecorder()
998+
errCh = make(chan error, 1)
999+
)
1000+
go func() {
1001+
url := fmt.Sprintf("/v1/health/service/web?dc=dc1&index=%d&wait=%s&%s", lastIndex, timeout, filterUrlPart)
1002+
req, err := http.NewRequest("GET", url, nil)
1003+
if err != nil {
1004+
errCh <- err
1005+
return
1006+
}
1007+
1008+
obj, err := a.srv.HealthServiceNodes(resp, req)
1009+
if err != nil {
1010+
errCh <- err
1011+
return
1012+
}
1013+
1014+
nodes := obj.(structs.CheckServiceNodes)
1015+
out = nodes
1016+
errCh <- nil
1017+
}()
1018+
1019+
time.Sleep(200 * time.Millisecond)
1020+
1021+
// Change the tags.
1022+
register(t, a, "web", "bar")
1023+
1024+
if err := <-errCh; err != nil {
1025+
require.NoError(t, err)
1026+
}
1027+
1028+
require.Len(t, out, 0)
1029+
require.Equal(t, tc.queryBackend, resp.Header().Get("X-Consul-Query-Backend"))
1030+
})
1031+
})
1032+
}
1033+
}
1034+
9051035
func TestHealthServiceNodes_NodeMetaFilter(t *testing.T) {
9061036
if testing.Short() {
9071037
t.Skip("too slow for testing.Short")

Diff for: agent/rpcclient/health/view.go

+4-3
Original file line numberDiff line numberDiff line change
@@ -80,11 +80,12 @@ func (s *healthView) Update(events []*pbsubscribe.Event) error {
8080
return errors.New("check service node was unexpectedly nil")
8181
}
8282
passed, err := s.filter.Evaluate(*csn)
83-
switch {
84-
case err != nil:
83+
if err != nil {
8584
return err
86-
case passed:
85+
} else if passed {
8786
s.state[id] = *csn
87+
} else {
88+
delete(s.state, id)
8889
}
8990

9091
case pbsubscribe.CatalogOp_Deregister:

Diff for: agent/structs/structs.go

+2
Original file line numberDiff line numberDiff line change
@@ -372,6 +372,8 @@ const (
372372
QueryBackendStreaming
373373
)
374374

375+
func (q QueryBackend) GoString() string { return q.String() }
376+
375377
func (q QueryBackend) String() string {
376378
switch q {
377379
case QueryBackendBlocking:

0 commit comments

Comments
 (0)