etcd/tests/e2e/http_health_check_test.go
Siyuan Zhang 12b640523a etcdserver: add linearizable_read check to readyz.
Signed-off-by: Siyuan Zhang <sizhang@google.com>
2023-11-21 11:13:20 -08:00

443 lines
14 KiB
Go

// Copyright 2023 The etcd Authors
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
//go:build !cluster_proxy
package e2e
import (
"context"
"fmt"
"io"
"net/http"
"os"
"path"
"strings"
"testing"
"time"
"github.com/stretchr/testify/require"
"go.etcd.io/etcd/api/v3/etcdserverpb"
"go.etcd.io/etcd/server/v3/storage/mvcc/testutil"
"go.etcd.io/etcd/tests/v3/framework/config"
"go.etcd.io/etcd/tests/v3/framework/e2e"
"go.etcd.io/etcd/tests/v3/framework/testutils"
)
const (
healthCheckTimeout = 2 * time.Second
putCommandTimeout = 200 * time.Millisecond
)
type healthCheckConfig struct {
url string
expectedStatusCode int
expectedTimeoutError bool
expectedRespSubStrings []string
}
type injectFailure func(ctx context.Context, t *testing.T, clus *e2e.EtcdProcessCluster, duration time.Duration)
func TestHTTPHealthHandler(t *testing.T) {
e2e.BeforeTest(t)
client := &http.Client{}
tcs := []struct {
name string
injectFailure injectFailure
clusterOptions []e2e.EPClusterOption
healthChecks []healthCheckConfig
}{
{
name: "no failures", // happy case
clusterOptions: []e2e.EPClusterOption{e2e.WithClusterSize(1)},
healthChecks: []healthCheckConfig{
{
url: "/health",
expectedStatusCode: http.StatusOK,
},
},
},
{
name: "activated no space alarm",
injectFailure: triggerNoSpaceAlarm,
clusterOptions: []e2e.EPClusterOption{e2e.WithClusterSize(1), e2e.WithQuotaBackendBytes(int64(13 * os.Getpagesize()))},
healthChecks: []healthCheckConfig{
{
url: "/health",
expectedStatusCode: http.StatusServiceUnavailable,
},
{
url: "/health?exclude=NOSPACE",
expectedStatusCode: http.StatusOK,
},
},
},
{
name: "overloaded server slow apply",
injectFailure: triggerSlowApply,
clusterOptions: []e2e.EPClusterOption{e2e.WithClusterSize(3), e2e.WithGoFailEnabled(true)},
healthChecks: []healthCheckConfig{
{
url: "/health?serializable=true",
expectedStatusCode: http.StatusOK,
},
{
url: "/health?serializable=false",
expectedTimeoutError: true,
},
},
},
{
name: "network partitioned",
injectFailure: blackhole,
clusterOptions: []e2e.EPClusterOption{e2e.WithClusterSize(3), e2e.WithIsPeerTLS(true), e2e.WithPeerProxy(true)},
healthChecks: []healthCheckConfig{
{
url: "/health?serializable=true",
expectedStatusCode: http.StatusOK,
},
{
url: "/health?serializable=false",
expectedTimeoutError: true,
expectedStatusCode: http.StatusServiceUnavailable,
// old leader may return "etcdserver: leader changed" error with 503 in ReadIndex leaderChangedNotifier
},
},
},
{
name: "raft loop deadlock",
injectFailure: triggerRaftLoopDeadLock,
clusterOptions: []e2e.EPClusterOption{e2e.WithClusterSize(1), e2e.WithGoFailEnabled(true)},
healthChecks: []healthCheckConfig{
{
// current kubeadm etcd liveness check failed to detect raft loop deadlock in steady state
// ref. https://github.com/kubernetes/kubernetes/blob/master/cmd/kubeadm/app/phases/etcd/local.go#L225-L226
// current liveness probe depends on the etcd /health check has a flaw that new /livez check should resolve.
url: "/health?serializable=true",
expectedStatusCode: http.StatusOK,
},
{
url: "/health?serializable=false",
expectedTimeoutError: true,
},
},
},
// verify that auth enabled serializable read must go through mvcc
{
name: "slow buffer write back with auth enabled",
injectFailure: triggerSlowBufferWriteBackWithAuth,
clusterOptions: []e2e.EPClusterOption{e2e.WithClusterSize(1), e2e.WithGoFailEnabled(true)},
healthChecks: []healthCheckConfig{
{
url: "/health?serializable=true",
expectedTimeoutError: true,
},
},
},
}
for _, tc := range tcs {
t.Run(tc.name, func(t *testing.T) {
ctx, cancel := context.WithTimeout(context.Background(), 20*time.Second)
defer cancel()
clus, err := e2e.NewEtcdProcessCluster(ctx, t, tc.clusterOptions...)
require.NoError(t, err)
defer clus.Close()
testutils.ExecuteUntil(ctx, t, func() {
if tc.injectFailure != nil {
// guaranteed that failure point is active until all the health checks timeout.
duration := time.Duration(len(tc.healthChecks)+1) * healthCheckTimeout
tc.injectFailure(ctx, t, clus, duration)
}
for _, hc := range tc.healthChecks {
requestURL := clus.Procs[0].EndpointsHTTP()[0] + hc.url
t.Logf("health check URL is %s", requestURL)
doHealthCheckAndVerify(t, client, requestURL, hc.expectedTimeoutError, hc.expectedStatusCode, hc.expectedRespSubStrings)
}
})
})
}
}
var (
defaultHealthCheckConfigs = []healthCheckConfig{
{
url: "/livez",
expectedStatusCode: http.StatusOK,
expectedRespSubStrings: []string{`ok`},
},
{
url: "/readyz",
expectedStatusCode: http.StatusOK,
expectedRespSubStrings: []string{`ok`},
},
{
url: "/livez?verbose=true",
expectedStatusCode: http.StatusOK,
expectedRespSubStrings: []string{`[+]serializable_read ok`},
},
{
url: "/readyz?verbose=true",
expectedStatusCode: http.StatusOK,
expectedRespSubStrings: []string{
`[+]serializable_read ok`,
`[+]data_corruption ok`,
},
},
}
)
func TestHTTPLivezReadyzHandler(t *testing.T) {
e2e.BeforeTest(t)
client := &http.Client{}
tcs := []struct {
name string
injectFailure injectFailure
clusterOptions []e2e.EPClusterOption
healthChecks []healthCheckConfig
}{
{
name: "no failures", // happy case
clusterOptions: []e2e.EPClusterOption{e2e.WithClusterSize(1)},
healthChecks: defaultHealthCheckConfigs,
},
{
name: "activated no space alarm",
injectFailure: triggerNoSpaceAlarm,
clusterOptions: []e2e.EPClusterOption{e2e.WithClusterSize(1), e2e.WithQuotaBackendBytes(int64(13 * os.Getpagesize()))},
healthChecks: defaultHealthCheckConfigs,
},
// Readiness is not an indicator of performance. Slow response is not covered by readiness.
// refer to https://tinyurl.com/livez-readyz-design-doc or https://github.com/etcd-io/etcd/issues/16007#issuecomment-1726541091 in case tinyurl is down.
{
name: "overloaded server slow apply",
injectFailure: triggerSlowApply,
clusterOptions: []e2e.EPClusterOption{e2e.WithClusterSize(3), e2e.WithGoFailEnabled(true)},
// TODO expected behavior of readyz check should be 200 after ReadIndex check is implemented to replace linearizable read.
healthChecks: []healthCheckConfig{
{
url: "/livez",
expectedStatusCode: http.StatusOK,
},
{
url: "/readyz",
expectedTimeoutError: true,
expectedStatusCode: http.StatusServiceUnavailable,
},
},
},
{
name: "network partitioned",
injectFailure: blackhole,
clusterOptions: []e2e.EPClusterOption{e2e.WithClusterSize(3), e2e.WithIsPeerTLS(true), e2e.WithPeerProxy(true)},
healthChecks: []healthCheckConfig{
{
url: "/livez",
expectedStatusCode: http.StatusOK,
},
{
url: "/readyz",
expectedTimeoutError: true,
expectedStatusCode: http.StatusServiceUnavailable,
expectedRespSubStrings: []string{
`[-]linearizable_read failed: etcdserver: leader changed`,
},
},
},
},
{
name: "raft loop deadlock",
injectFailure: triggerRaftLoopDeadLock,
clusterOptions: []e2e.EPClusterOption{e2e.WithClusterSize(1), e2e.WithGoFailEnabled(true)},
// TODO expected behavior of livez check should be 503 or timeout after RaftLoopDeadLock check is implemented.
healthChecks: []healthCheckConfig{
{
url: "/livez",
expectedStatusCode: http.StatusOK,
},
{
url: "/readyz",
expectedTimeoutError: true,
expectedStatusCode: http.StatusServiceUnavailable,
},
},
},
// verify that auth enabled serializable read must go through mvcc
{
name: "slow buffer write back with auth enabled",
injectFailure: triggerSlowBufferWriteBackWithAuth,
clusterOptions: []e2e.EPClusterOption{e2e.WithClusterSize(1), e2e.WithGoFailEnabled(true)},
healthChecks: []healthCheckConfig{
{
url: "/livez",
expectedTimeoutError: true,
},
{
url: "/readyz",
expectedTimeoutError: true,
},
},
},
{
name: "corrupt",
injectFailure: triggerCorrupt,
clusterOptions: []e2e.EPClusterOption{e2e.WithClusterSize(3), e2e.WithCorruptCheckTime(time.Second)},
healthChecks: []healthCheckConfig{
{
url: "/livez?verbose=true",
expectedStatusCode: http.StatusOK,
expectedRespSubStrings: []string{`[+]serializable_read ok`},
},
{
url: "/readyz",
expectedStatusCode: http.StatusServiceUnavailable,
expectedRespSubStrings: []string{
`[+]serializable_read ok`,
`[-]data_corruption failed: alarm activated: CORRUPT`,
},
},
},
},
}
for _, tc := range tcs {
t.Run(tc.name, func(t *testing.T) {
ctx, cancel := context.WithTimeout(context.Background(), 20*time.Second)
defer cancel()
clus, err := e2e.NewEtcdProcessCluster(ctx, t, tc.clusterOptions...)
require.NoError(t, err)
defer clus.Close()
testutils.ExecuteUntil(ctx, t, func() {
if tc.injectFailure != nil {
// guaranteed that failure point is active until all the health checks timeout.
duration := time.Duration(len(tc.healthChecks)+1) * healthCheckTimeout
tc.injectFailure(ctx, t, clus, duration)
}
for _, hc := range tc.healthChecks {
requestURL := clus.Procs[0].EndpointsHTTP()[0] + hc.url
t.Logf("health check URL is %s", requestURL)
doHealthCheckAndVerify(t, client, requestURL, hc.expectedTimeoutError, hc.expectedStatusCode, hc.expectedRespSubStrings)
}
})
})
}
}
func doHealthCheckAndVerify(t *testing.T, client *http.Client, url string, expectTimeoutError bool, expectStatusCode int, expectRespSubStrings []string) {
ctx, cancel := context.WithTimeout(context.Background(), healthCheckTimeout)
req, err := http.NewRequestWithContext(ctx, "GET", url, nil)
require.NoErrorf(t, err, "failed to creat request %+v", err)
resp, herr := client.Do(req)
cancel()
if expectTimeoutError {
if herr != nil && strings.Contains(herr.Error(), context.DeadlineExceeded.Error()) {
return
}
}
require.NoErrorf(t, herr, "failed to get response %+v", err)
defer resp.Body.Close()
body, err := io.ReadAll(resp.Body)
resp.Body.Close()
require.NoErrorf(t, err, "failed to read response %+v", err)
t.Logf("health check response body is:\n%s", body)
require.Equal(t, expectStatusCode, resp.StatusCode)
for _, expectRespSubString := range expectRespSubStrings {
require.Contains(t, string(body), expectRespSubString)
}
}
func triggerNoSpaceAlarm(ctx context.Context, t *testing.T, clus *e2e.EtcdProcessCluster, _ time.Duration) {
buf := strings.Repeat("b", os.Getpagesize())
etcdctl := clus.Etcdctl()
for {
if err := etcdctl.Put(ctx, "foo", buf, config.PutOptions{}); err != nil {
if !strings.Contains(err.Error(), "etcdserver: mvcc: database space exceeded") {
t.Fatal(err)
}
break
}
}
}
func triggerSlowApply(ctx context.Context, t *testing.T, clus *e2e.EtcdProcessCluster, duration time.Duration) {
// the following proposal will be blocked at applying stage
// because when apply index < committed index, linearizable read would time out.
require.NoError(t, clus.Procs[0].Failpoints().SetupHTTP(ctx, "beforeApplyOneEntryNormal", fmt.Sprintf(`sleep("%s")`, duration)))
require.NoError(t, clus.Procs[1].Etcdctl().Put(ctx, "foo", "bar", config.PutOptions{}))
}
func blackhole(_ context.Context, t *testing.T, clus *e2e.EtcdProcessCluster, _ time.Duration) {
member := clus.Procs[0]
proxy := member.PeerProxy()
t.Logf("Blackholing traffic from and to member %q", member.Config().Name)
proxy.BlackholeTx()
proxy.BlackholeRx()
}
func triggerRaftLoopDeadLock(ctx context.Context, t *testing.T, clus *e2e.EtcdProcessCluster, duration time.Duration) {
require.NoError(t, clus.Procs[0].Failpoints().SetupHTTP(ctx, "raftBeforeSave", fmt.Sprintf(`sleep("%s")`, duration)))
clus.Procs[0].Etcdctl().Put(context.Background(), "foo", "bar", config.PutOptions{Timeout: putCommandTimeout})
}
func triggerSlowBufferWriteBackWithAuth(ctx context.Context, t *testing.T, clus *e2e.EtcdProcessCluster, duration time.Duration) {
etcdctl := clus.Etcdctl()
_, err := etcdctl.UserAdd(ctx, "root", "root", config.UserAddOptions{})
require.NoError(t, err)
_, err = etcdctl.UserGrantRole(ctx, "root", "root")
require.NoError(t, err)
require.NoError(t, etcdctl.AuthEnable(ctx))
require.NoError(t, clus.Procs[0].Failpoints().SetupHTTP(ctx, "beforeWritebackBuf", fmt.Sprintf(`sleep("%s")`, duration)))
clus.Procs[0].Etcdctl(e2e.WithAuth("root", "root")).Put(context.Background(), "foo", "bar", config.PutOptions{Timeout: putCommandTimeout})
}
func triggerCorrupt(ctx context.Context, t *testing.T, clus *e2e.EtcdProcessCluster, _ time.Duration) {
etcdctl := clus.Procs[0].Etcdctl()
for i := 0; i < 10; i++ {
err := etcdctl.Put(ctx, "foo", "bar", config.PutOptions{})
require.NoError(t, err)
}
err := clus.Procs[0].Kill()
require.NoError(t, err)
err = clus.Procs[0].Wait(ctx)
require.NoError(t, err)
err = testutil.CorruptBBolt(path.Join(clus.Procs[0].Config().DataDirPath, "member", "snap", "db"))
require.NoError(t, err)
err = clus.Procs[0].Start(ctx)
for {
time.Sleep(time.Second)
select {
case <-ctx.Done():
require.NoError(t, err)
default:
}
response, err := etcdctl.AlarmList(ctx)
if err != nil {
continue
}
if len(response.Alarms) == 0 {
continue
}
require.Len(t, response.Alarms, 1)
if response.Alarms[0].Alarm == etcdserverpb.AlarmType_CORRUPT {
break
}
}
}