etcd/tests/e2e/watch_test.go
Wei Fu db701a5fb7 *: update tests for watch API when compact on tombstone revision
Signed-off-by: Wei Fu <fuweid89@gmail.com>
(cherry picked from commit ee33652775842b96d1c0ab601ec3002078998c2a)
Signed-off-by: Wei Fu <fuweid89@gmail.com>
2024-08-20 18:22:09 +08:00

485 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.
// These tests are performance sensitive, addition of cluster proxy makes them unstable.
//go:build !cluster_proxy
package e2e
import (
"context"
"fmt"
"strings"
"sync"
"testing"
"time"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"go.etcd.io/etcd/api/v3/etcdserverpb"
"go.etcd.io/etcd/api/v3/mvccpb"
v3rpc "go.etcd.io/etcd/api/v3/v3rpc/rpctypes"
clientv3 "go.etcd.io/etcd/client/v3"
"go.etcd.io/etcd/tests/v3/framework/e2e"
"golang.org/x/sync/errgroup"
)
const (
watchResponsePeriod = 100 * time.Millisecond
watchTestDuration = 5 * time.Second
readLoadConcurrency = 10
)
type testCase struct {
name string
config e2e.EtcdProcessClusterConfig
maxWatchDelay time.Duration
dbSizeBytes int
}
const (
Kilo = 1000
Mega = 1000 * Kilo
)
// 10 MB is not a bottleneck of grpc server, but filling up etcd with data.
// Keeping it lower so tests don't take too long.
// If we implement reuse of db we could increase the dbSize.
var tcs = []testCase{
{
name: "NoTLS",
config: e2e.EtcdProcessClusterConfig{ClusterSize: 1},
maxWatchDelay: 150 * time.Millisecond,
dbSizeBytes: 5 * Mega,
},
{
name: "TLS",
config: e2e.EtcdProcessClusterConfig{ClusterSize: 1, IsClientAutoTLS: true, ClientTLS: e2e.ClientTLS},
maxWatchDelay: 150 * time.Millisecond,
dbSizeBytes: 5 * Mega,
},
{
name: "SeparateHttpNoTLS",
config: e2e.EtcdProcessClusterConfig{ClusterSize: 1, ClientHttpSeparate: true},
maxWatchDelay: 150 * time.Millisecond,
dbSizeBytes: 5 * Mega,
},
{
name: "SeparateHttpTLS",
config: e2e.EtcdProcessClusterConfig{ClusterSize: 1, IsClientAutoTLS: true, ClientTLS: e2e.ClientTLS, ClientHttpSeparate: true},
maxWatchDelay: 150 * time.Millisecond,
dbSizeBytes: 5 * Mega,
},
}
func TestWatchDelayForPeriodicProgressNotification(t *testing.T) {
e2e.BeforeTest(t)
for _, tc := range tcs {
tc := tc
tc.config.WatchProcessNotifyInterval = watchResponsePeriod
t.Run(tc.name, func(t *testing.T) {
clus, err := e2e.NewEtcdProcessCluster(t, &tc.config)
require.NoError(t, err)
defer clus.Close()
c := newClient(t, clus.EndpointsV3(), tc.config.ClientTLS, tc.config.IsClientAutoTLS)
require.NoError(t, fillEtcdWithData(context.Background(), c, tc.dbSizeBytes))
ctx, cancel := context.WithTimeout(context.Background(), watchTestDuration)
defer cancel()
g := errgroup.Group{}
continuouslyExecuteGetAll(ctx, t, &g, c)
validateWatchDelay(t, c.Watch(ctx, "fake-key", clientv3.WithProgressNotify()), tc.maxWatchDelay)
require.NoError(t, g.Wait())
})
}
}
func TestWatchDelayForManualProgressNotification(t *testing.T) {
e2e.BeforeTest(t)
for _, tc := range tcs {
t.Run(tc.name, func(t *testing.T) {
clus, err := e2e.NewEtcdProcessCluster(t, &tc.config)
require.NoError(t, err)
defer clus.Close()
c := newClient(t, clus.EndpointsV3(), tc.config.ClientTLS, tc.config.IsClientAutoTLS)
require.NoError(t, fillEtcdWithData(context.Background(), c, tc.dbSizeBytes))
ctx, cancel := context.WithTimeout(context.Background(), watchTestDuration)
defer cancel()
g := errgroup.Group{}
continuouslyExecuteGetAll(ctx, t, &g, c)
g.Go(func() error {
for {
err := c.RequestProgress(ctx)
if err != nil {
if strings.Contains(err.Error(), "context deadline exceeded") {
return nil
}
return err
}
time.Sleep(watchResponsePeriod)
}
})
validateWatchDelay(t, c.Watch(ctx, "fake-key"), tc.maxWatchDelay)
require.NoError(t, g.Wait())
})
}
}
func TestWatchDelayForEvent(t *testing.T) {
e2e.BeforeTest(t)
for _, tc := range tcs {
t.Run(tc.name, func(t *testing.T) {
clus, err := e2e.NewEtcdProcessCluster(t, &tc.config)
require.NoError(t, err)
defer clus.Close()
c := newClient(t, clus.EndpointsV3(), tc.config.ClientTLS, tc.config.IsClientAutoTLS)
require.NoError(t, fillEtcdWithData(context.Background(), c, tc.dbSizeBytes))
ctx, cancel := context.WithTimeout(context.Background(), watchTestDuration)
defer cancel()
g := errgroup.Group{}
g.Go(func() error {
i := 0
for {
_, err := c.Put(ctx, "key", fmt.Sprintf("%d", i))
if err != nil {
if strings.Contains(err.Error(), "context deadline exceeded") {
return nil
}
return err
}
time.Sleep(watchResponsePeriod)
}
})
continuouslyExecuteGetAll(ctx, t, &g, c)
validateWatchDelay(t, c.Watch(ctx, "key"), tc.maxWatchDelay)
require.NoError(t, g.Wait())
})
}
}
func validateWatchDelay(t *testing.T, watch clientv3.WatchChan, maxWatchDelay time.Duration) {
start := time.Now()
var maxDelay time.Duration
for range watch {
sinceLast := time.Since(start)
if sinceLast > watchResponsePeriod+maxWatchDelay {
t.Errorf("Unexpected watch response delayed over allowed threshold %s, delay: %s", maxWatchDelay, sinceLast-watchResponsePeriod)
} else {
t.Logf("Got watch response, since last: %s", sinceLast)
}
if sinceLast > maxDelay {
maxDelay = sinceLast
}
start = time.Now()
}
sinceLast := time.Since(start)
if sinceLast > maxDelay && sinceLast > watchResponsePeriod+maxWatchDelay {
t.Errorf("Unexpected watch response delayed over allowed threshold %s, delay: unknown", maxWatchDelay)
t.Errorf("Test finished while in middle of delayed response, measured delay: %s", sinceLast-watchResponsePeriod)
t.Logf("Please increase the test duration to measure delay")
} else {
t.Logf("Max delay: %s", maxDelay-watchResponsePeriod)
}
}
func continuouslyExecuteGetAll(ctx context.Context, t *testing.T, g *errgroup.Group, c *clientv3.Client) {
mux := sync.RWMutex{}
size := 0
for i := 0; i < readLoadConcurrency; i++ {
g.Go(func() error {
for {
resp, err := c.Get(ctx, "", clientv3.WithPrefix())
if err != nil {
if strings.Contains(err.Error(), "context deadline exceeded") {
return nil
}
return err
}
respSize := 0
for _, kv := range resp.Kvs {
respSize += kv.Size()
}
mux.Lock()
size += respSize
mux.Unlock()
}
})
}
g.Go(func() error {
lastSize := size
for range time.Tick(time.Second) {
select {
case <-ctx.Done():
return nil
default:
}
mux.RLock()
t.Logf("Generating read load around %.1f MB/s", float64(size-lastSize)/1000/1000)
lastSize = size
mux.RUnlock()
}
return nil
})
}
// TestDeleteEventDrop_Issue18089 is an e2e test to reproduce the issue reported in: https://github.com/etcd-io/etcd/issues/18089
//
// The goal is to reproduce a DELETE event being dropped in a watch after a compaction
// occurs on the revision where the deletion took place. In order to reproduce this, we
// perform the following sequence (steps for reproduction thanks to @ahrtr):
// - PUT k v2 (assume returned revision = r2)
// - PUT k v3 (assume returned revision = r3)
// - PUT k v4 (assume returned revision = r4)
// - DELETE k (assume returned revision = r5)
// - PUT k v6 (assume returned revision = r6)
// - COMPACT r5
// - WATCH rev=r5
//
// We should get the DELETE event (r5) followed by the PUT event (r6).
func TestDeleteEventDrop_Issue18089(t *testing.T) {
e2e.BeforeTest(t)
cfg := e2e.EtcdProcessClusterConfig{
ClusterSize: 1,
IsClientAutoTLS: true,
ClientTLS: e2e.ClientTLS,
}
clus, err := e2e.NewEtcdProcessCluster(t, &cfg)
require.NoError(t, err)
defer clus.Close()
c := newClient(t, clus.EndpointsGRPC(), cfg.ClientTLS, cfg.IsClientAutoTLS)
defer c.Close()
ctx := context.Background()
const (
key = "k"
v2 = "v2"
v3 = "v3"
v4 = "v4"
v6 = "v6"
)
t.Logf("PUT key=%s, val=%s", key, v2)
_, err = c.KV.Put(ctx, key, v2)
require.NoError(t, err)
t.Logf("PUT key=%s, val=%s", key, v3)
_, err = c.KV.Put(ctx, key, v3)
require.NoError(t, err)
t.Logf("PUT key=%s, val=%s", key, v4)
_, err = c.KV.Put(ctx, key, v4)
require.NoError(t, err)
t.Logf("DELTE key=%s", key)
deleteResp, err := c.KV.Delete(ctx, key)
require.NoError(t, err)
t.Logf("PUT key=%s, val=%s", key, v6)
_, err = c.KV.Put(ctx, key, v6)
require.NoError(t, err)
t.Logf("COMPACT rev=%d", deleteResp.Header.Revision)
_, err = c.KV.Compact(ctx, deleteResp.Header.Revision, clientv3.WithCompactPhysical())
require.NoError(t, err)
watchChan := c.Watch(ctx, key, clientv3.WithRev(deleteResp.Header.Revision))
select {
case watchResp := <-watchChan:
require.Len(t, watchResp.Events, 2)
require.Equal(t, mvccpb.DELETE, watchResp.Events[0].Type)
deletedKey := string(watchResp.Events[0].Kv.Key)
require.Equal(t, key, deletedKey)
require.Equal(t, mvccpb.PUT, watchResp.Events[1].Type)
updatedKey := string(watchResp.Events[1].Kv.Key)
require.Equal(t, key, updatedKey)
require.Equal(t, v6, string(watchResp.Events[1].Kv.Value))
case <-time.After(100 * time.Millisecond):
// we care only about the first response, but have an
// escape hatch in case the watch response is delayed.
t.Fatal("timed out getting watch response")
}
}
func TestStartWatcherFromCompactedRevision(t *testing.T) {
t.Run("compaction on tombstone revision", func(t *testing.T) {
testStartWatcherFromCompactedRevision(t, true)
})
t.Run("compaction on normal revision", func(t *testing.T) {
testStartWatcherFromCompactedRevision(t, false)
})
}
func testStartWatcherFromCompactedRevision(t *testing.T, performCompactOnTombstone bool) {
e2e.BeforeTest(t)
cfg := e2e.EtcdProcessClusterConfig{
ClusterSize: 1,
IsClientAutoTLS: true,
ClientTLS: e2e.ClientTLS,
}
clus, err := e2e.NewEtcdProcessCluster(t, &cfg)
require.NoError(t, err)
defer clus.Close()
c := newClient(t, clus.EndpointsGRPC(), cfg.ClientTLS, cfg.IsClientAutoTLS)
defer c.Close()
ctx := context.Background()
key := "foo"
totalRev := 100
type valueEvent struct {
value string
typ mvccpb.Event_EventType
}
var (
// requestedValues records all requested change
requestedValues = make([]valueEvent, 0)
// revisionChan sends each compacted revision via this channel
compactionRevChan = make(chan int64)
// compactionStep means that client performs a compaction on every 7 operations
compactionStep = 7
)
// This goroutine will submit changes on $key $totalRev times. It will
// perform compaction after every $compactedAfterChanges changes.
// Except for first time, the watcher always receives the compacted
// revision as start.
go func() {
defer close(compactionRevChan)
lastRevision := int64(1)
compactionRevChan <- lastRevision
for vi := 1; vi <= totalRev; vi++ {
var respHeader *etcdserverpb.ResponseHeader
if vi%compactionStep == 0 && performCompactOnTombstone {
t.Logf("DELETE key=%s", key)
resp, derr := c.KV.Delete(ctx, key)
require.NoError(t, derr)
respHeader = resp.Header
requestedValues = append(requestedValues, valueEvent{value: "", typ: mvccpb.DELETE})
} else {
value := fmt.Sprintf("%d", vi)
t.Logf("PUT key=%s, val=%s", key, value)
resp, perr := c.KV.Put(ctx, key, value)
require.NoError(t, perr)
respHeader = resp.Header
requestedValues = append(requestedValues, valueEvent{value: value, typ: mvccpb.PUT})
}
lastRevision = respHeader.Revision
if vi%compactionStep == 0 {
compactionRevChan <- lastRevision
t.Logf("COMPACT rev=%d", lastRevision)
_, err = c.KV.Compact(ctx, lastRevision, clientv3.WithCompactPhysical())
require.NoError(t, err)
}
}
}()
receivedEvents := make([]*clientv3.Event, 0)
fromCompactedRev := false
for fromRev := range compactionRevChan {
watchChan := c.Watch(ctx, key, clientv3.WithRev(fromRev))
prevEventCount := len(receivedEvents)
// firstReceived represents this is first watch response.
// Just in case that ETCD sends event one by one.
firstReceived := true
t.Logf("Start to watch key %s starting from revision %d", key, fromRev)
watchLoop:
for {
currentEventCount := len(receivedEvents)
if currentEventCount-prevEventCount == compactionStep || currentEventCount == totalRev {
break
}
select {
case watchResp := <-watchChan:
t.Logf("Receive the number of events: %d", len(watchResp.Events))
for i := range watchResp.Events {
ev := watchResp.Events[i]
// If the $fromRev is the compacted revision,
// the first event should be the same as the last event receives in last watch response.
if firstReceived && fromCompactedRev {
firstReceived = false
last := receivedEvents[prevEventCount-1]
assert.Equal(t, last.Type, ev.Type,
"last received event type %s, but got event type %s", last.Type, ev.Type)
assert.Equal(t, string(last.Kv.Key), string(ev.Kv.Key),
"last received event key %s, but got event key %s", string(last.Kv.Key), string(ev.Kv.Key))
assert.Equal(t, string(last.Kv.Value), string(ev.Kv.Value),
"last received event value %s, but got event value %s", string(last.Kv.Value), string(ev.Kv.Value))
continue
}
receivedEvents = append(receivedEvents, ev)
}
if len(watchResp.Events) == 0 {
require.Equal(t, v3rpc.ErrCompacted, watchResp.Err())
break watchLoop
}
case <-time.After(10 * time.Second):
t.Fatal("timed out getting watch response")
}
}
fromCompactedRev = true
}
t.Logf("Received total number of events: %d", len(receivedEvents))
require.Len(t, requestedValues, totalRev)
require.Len(t, receivedEvents, totalRev, "should receive %d events", totalRev)
for idx, expected := range requestedValues {
ev := receivedEvents[idx]
require.Equal(t, expected.typ, ev.Type, "#%d expected event %s", idx, expected.typ)
updatedKey := string(ev.Kv.Key)
require.Equal(t, key, updatedKey)
if expected.typ == mvccpb.PUT {
updatedValue := string(ev.Kv.Value)
require.Equal(t, expected.value, updatedValue)
}
}
}