
When hashes don't match, there could be some nodes falling behind and the compact request can then error with 'future revision compact'.
229 lines
6.4 KiB
Go
229 lines
6.4 KiB
Go
// Copyright 2015 CoreOS, Inc.
|
|
//
|
|
// 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.
|
|
|
|
package main
|
|
|
|
import (
|
|
"sync"
|
|
"time"
|
|
)
|
|
|
|
type tester struct {
|
|
failures []failure
|
|
cluster *cluster
|
|
limit int
|
|
|
|
status Status
|
|
}
|
|
|
|
func (tt *tester) runLoop() {
|
|
tt.status.Since = time.Now()
|
|
tt.status.RoundLimit = tt.limit
|
|
tt.status.cluster = tt.cluster
|
|
for _, f := range tt.failures {
|
|
tt.status.Failures = append(tt.status.Failures, f.Desc())
|
|
}
|
|
for i := 0; i < tt.limit; i++ {
|
|
tt.status.setRound(i)
|
|
roundTotalCounter.Inc()
|
|
|
|
var (
|
|
currentRevision int64
|
|
success bool
|
|
)
|
|
for j, f := range tt.failures {
|
|
caseTotalCounter.WithLabelValues(f.Desc()).Inc()
|
|
|
|
tt.status.setCase(j)
|
|
|
|
if err := tt.cluster.WaitHealth(); err != nil {
|
|
plog.Printf("[round#%d case#%d] wait full health error: %v", i, j, err)
|
|
if err := tt.cleanup(i, j); err != nil {
|
|
plog.Printf("[round#%d case#%d] cleanup error: %v", i, j, err)
|
|
return
|
|
}
|
|
break
|
|
}
|
|
plog.Printf("[round#%d case#%d] start failure %s", i, j, f.Desc())
|
|
|
|
plog.Printf("[round#%d case#%d] start injecting failure...", i, j)
|
|
if err := f.Inject(tt.cluster, i); err != nil {
|
|
plog.Printf("[round#%d case#%d] injection error: %v", i, j, err)
|
|
if err := tt.cleanup(i, j); err != nil {
|
|
plog.Printf("[round#%d case#%d] cleanup error: %v", i, j, err)
|
|
return
|
|
}
|
|
break
|
|
}
|
|
plog.Printf("[round#%d case#%d] injected failure", i, j)
|
|
|
|
plog.Printf("[round#%d case#%d] start recovering failure...", i, j)
|
|
if err := f.Recover(tt.cluster, i); err != nil {
|
|
plog.Printf("[round#%d case#%d] recovery error: %v", i, j, err)
|
|
if err := tt.cleanup(i, j); err != nil {
|
|
plog.Printf("[round#%d case#%d] cleanup error: %v", i, j, err)
|
|
return
|
|
}
|
|
break
|
|
}
|
|
plog.Printf("[round#%d case#%d] recovered failure", i, j)
|
|
|
|
if tt.cluster.v2Only {
|
|
plog.Printf("[round#%d case#%d] succeed!", i, j)
|
|
continue
|
|
}
|
|
|
|
plog.Printf("[round#%d case#%d] canceling the stressers...", i, j)
|
|
for _, s := range tt.cluster.Stressers {
|
|
s.Cancel()
|
|
}
|
|
plog.Printf("[round#%d case#%d] canceled stressers", i, j)
|
|
|
|
plog.Printf("[round#%d case#%d] checking current revisions...", i, j)
|
|
var (
|
|
revs map[string]int64
|
|
hashes map[string]int64
|
|
rerr error
|
|
ok bool
|
|
)
|
|
for k := 0; k < 5; k++ {
|
|
time.Sleep(time.Second)
|
|
|
|
revs, hashes, rerr = tt.cluster.getRevisionHash()
|
|
if rerr != nil {
|
|
plog.Printf("[round#%d case#%d.%d] failed to get current revisions (%v)", i, j, k, rerr)
|
|
continue
|
|
}
|
|
if currentRevision, ok = getSameValue(revs); ok {
|
|
break
|
|
}
|
|
|
|
plog.Printf("[round#%d case#%d.%d] inconsistent current revisions %+v", i, j, k, revs)
|
|
}
|
|
if !ok || rerr != nil {
|
|
plog.Printf("[round#%d case#%d] checking current revisions failed [revisions: %v]", i, j, revs)
|
|
if err := tt.cleanup(i, j); err != nil {
|
|
plog.Printf("[round#%d case#%d] cleanup error: %v", i, j, err)
|
|
return
|
|
}
|
|
break
|
|
}
|
|
plog.Printf("[round#%d case#%d] all members are consistent with current revisions [revisions: %v]", i, j, revs)
|
|
|
|
plog.Printf("[round#%d case#%d] checking current storage hashes...", i, j)
|
|
if _, ok = getSameValue(hashes); !ok {
|
|
plog.Printf("[round#%d case#%d] checking current storage hashes failed [hashes: %v]", i, j, hashes)
|
|
if err := tt.cleanup(i, j); err != nil {
|
|
plog.Printf("[round#%d case#%d] cleanup error: %v", i, j, err)
|
|
return
|
|
}
|
|
break
|
|
}
|
|
plog.Printf("[round#%d case#%d] all members are consistent with storage hashes", i, j)
|
|
|
|
plog.Printf("[round#%d case#%d] restarting the stressers...", i, j)
|
|
for _, s := range tt.cluster.Stressers {
|
|
go s.Stress()
|
|
}
|
|
|
|
plog.Printf("[round#%d case#%d] succeed!", i, j)
|
|
success = true
|
|
}
|
|
|
|
if !success {
|
|
continue
|
|
}
|
|
revToCompact := max(0, currentRevision-10000)
|
|
plog.Printf("[round#%d] compacting storage at %d (current revision %d)", i, revToCompact, currentRevision)
|
|
if err := tt.cluster.compactKV(revToCompact); err != nil {
|
|
plog.Printf("[round#%d] compactKV error (%v)", i, err)
|
|
if err := tt.cleanup(i, 0); err != nil {
|
|
plog.Printf("[round#%d] cleanup error: %v", i, err)
|
|
return
|
|
}
|
|
continue
|
|
}
|
|
plog.Printf("[round#%d] compacted storage", i)
|
|
|
|
plog.Printf("[round#%d] check compaction at %d", i, revToCompact)
|
|
if err := tt.cluster.checkCompact(revToCompact); err != nil {
|
|
plog.Printf("[round#%d] checkCompact error (%v)", i, err)
|
|
if err := tt.cleanup(i, 0); err != nil {
|
|
plog.Printf("[round#%d] cleanup error: %v", i, err)
|
|
return
|
|
}
|
|
}
|
|
plog.Printf("[round#%d] confirmed compaction at %d", i, revToCompact)
|
|
|
|
if i > 0 && i%500 == 0 { // every 500 rounds
|
|
plog.Printf("[round#%d] canceling the stressers...", i)
|
|
for _, s := range tt.cluster.Stressers {
|
|
s.Cancel()
|
|
}
|
|
plog.Printf("[round#%d] canceled stressers", i)
|
|
|
|
plog.Printf("[round#%d] deframenting...", i)
|
|
if err := tt.cluster.defrag(); err != nil {
|
|
plog.Printf("[round#%d] defrag error (%v)", i, err)
|
|
if err := tt.cleanup(i, 0); err != nil {
|
|
plog.Printf("[round#%d] cleanup error: %v", i, err)
|
|
return
|
|
}
|
|
}
|
|
plog.Printf("[round#%d] deframented...", i)
|
|
|
|
plog.Printf("[round#%d] restarting the stressers...", i)
|
|
for _, s := range tt.cluster.Stressers {
|
|
go s.Stress()
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
func (tt *tester) cleanup(i, j int) error {
|
|
roundFailedTotalCounter.Inc()
|
|
caseFailedTotalCounter.WithLabelValues(tt.failures[j].Desc()).Inc()
|
|
|
|
plog.Printf("[round#%d case#%d] cleaning up...", i, j)
|
|
if err := tt.cluster.Cleanup(); err != nil {
|
|
return err
|
|
}
|
|
return tt.cluster.Bootstrap()
|
|
}
|
|
|
|
type Status struct {
|
|
Since time.Time
|
|
Failures []string
|
|
RoundLimit int
|
|
|
|
Cluster ClusterStatus
|
|
cluster *cluster
|
|
|
|
mu sync.Mutex // guards Round and Case
|
|
Round int
|
|
Case int
|
|
}
|
|
|
|
func (s *Status) setRound(r int) {
|
|
s.mu.Lock()
|
|
defer s.mu.Unlock()
|
|
s.Round = r
|
|
}
|
|
|
|
func (s *Status) setCase(c int) {
|
|
s.mu.Lock()
|
|
defer s.mu.Unlock()
|
|
s.Case = c
|
|
}
|