Merge pull request #4187 from prometheus/2972-remote-block-shutdown

Only give remote queues 1 minute to flush samples on shutdown.
This commit is contained in:
Tom Wilkie 2018-05-29 14:25:19 +01:00 committed by GitHub
commit 12a1a22a5a
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
6 changed files with 107 additions and 37 deletions

View file

@ -92,6 +92,7 @@ func main() {
webTimeout model.Duration
queryTimeout model.Duration
queryConcurrency int
RemoteFlushDeadline model.Duration
prometheusURL string
@ -160,6 +161,9 @@ func main() {
a.Flag("storage.tsdb.no-lockfile", "Do not create lockfile in data directory.").
Default("false").BoolVar(&cfg.tsdb.NoLockfile)
a.Flag("storage.remote.flush-deadline", "How long to wait flushing sample on shutdown or config reload.").
Default("1m").PlaceHolder("<duration>").SetValue(&cfg.RemoteFlushDeadline)
a.Flag("alertmanager.notification-queue-capacity", "The capacity of the queue for pending Alertmanager notifications.").
Default("10000").IntVar(&cfg.notifier.QueueCapacity)
@ -222,7 +226,7 @@ func main() {
var (
localStorage = &tsdb.ReadyStorage{}
remoteStorage = remote.NewStorage(log.With(logger, "component", "remote"), localStorage.StartTime)
remoteStorage = remote.NewStorage(log.With(logger, "component", "remote"), localStorage.StartTime, time.Duration(cfg.RemoteFlushDeadline))
fanoutStorage = storage.NewFanout(logger, localStorage, remoteStorage)
)

View file

@ -69,7 +69,7 @@ type recoverableError struct {
}
// Store sends a batch of samples to the HTTP endpoint.
func (c *Client) Store(req *prompb.WriteRequest) error {
func (c *Client) Store(ctx context.Context, req *prompb.WriteRequest) error {
data, err := proto.Marshal(req)
if err != nil {
return err
@ -85,6 +85,7 @@ func (c *Client) Store(req *prompb.WriteRequest) error {
httpReq.Header.Add("Content-Encoding", "snappy")
httpReq.Header.Set("Content-Type", "application/x-protobuf")
httpReq.Header.Set("X-Prometheus-Remote-Write-Version", "0.1.0")
httpReq = httpReq.WithContext(ctx)
ctx, cancel := context.WithTimeout(context.Background(), c.timeout)
defer cancel()

View file

@ -14,6 +14,7 @@
package remote
import (
"context"
"fmt"
"net/http"
"net/http/httptest"
@ -73,7 +74,7 @@ func TestStoreHTTPErrorHandling(t *testing.T) {
t.Fatal(err)
}
err = c.Store(&prompb.WriteRequest{})
err = c.Store(context.Background(), &prompb.WriteRequest{})
if !reflect.DeepEqual(err, test.err) {
t.Errorf("%d. Unexpected error; want %v, got %v", i, test.err, err)
}

View file

@ -14,8 +14,10 @@
package remote
import (
"context"
"math"
"sync"
"sync/atomic"
"time"
"golang.org/x/time/rate"
@ -129,7 +131,7 @@ func init() {
// external timeseries database.
type StorageClient interface {
// Store stores the given samples in the remote storage.
Store(*prompb.WriteRequest) error
Store(context.Context, *prompb.WriteRequest) error
// Name identifies the remote storage implementation.
Name() string
}
@ -139,6 +141,7 @@ type StorageClient interface {
type QueueManager struct {
logger log.Logger
flushDeadline time.Duration
cfg config.QueueConfig
externalLabels model.LabelSet
relabelConfigs []*config.RelabelConfig
@ -158,12 +161,13 @@ type QueueManager struct {
}
// NewQueueManager builds a new QueueManager.
func NewQueueManager(logger log.Logger, cfg config.QueueConfig, externalLabels model.LabelSet, relabelConfigs []*config.RelabelConfig, client StorageClient) *QueueManager {
func NewQueueManager(logger log.Logger, cfg config.QueueConfig, externalLabels model.LabelSet, relabelConfigs []*config.RelabelConfig, client StorageClient, flushDeadline time.Duration) *QueueManager {
if logger == nil {
logger = log.NewNopLogger()
}
t := &QueueManager{
logger: logger,
flushDeadline: flushDeadline,
cfg: cfg,
externalLabels: externalLabels,
relabelConfigs: relabelConfigs,
@ -255,7 +259,7 @@ func (t *QueueManager) Stop() {
t.shardsMtx.Lock()
defer t.shardsMtx.Unlock()
t.shards.stop()
t.shards.stop(t.flushDeadline)
level.Info(t.logger).Log("msg", "Remote storage stopped.")
}
@ -360,7 +364,7 @@ func (t *QueueManager) reshard(n int) {
t.shards = newShards
t.shardsMtx.Unlock()
oldShards.stop()
oldShards.stop(t.flushDeadline)
// We start the newShards after we have stopped (the therefore completely
// flushed) the oldShards, to guarantee we only every deliver samples in
@ -372,7 +376,9 @@ type shards struct {
qm *QueueManager
queues []chan *model.Sample
done chan struct{}
wg sync.WaitGroup
running int32
ctx context.Context
cancel context.CancelFunc
}
func (t *QueueManager) newShards(numShards int) *shards {
@ -380,12 +386,15 @@ func (t *QueueManager) newShards(numShards int) *shards {
for i := 0; i < numShards; i++ {
queues[i] = make(chan *model.Sample, t.cfg.Capacity)
}
ctx, cancel := context.WithCancel(context.Background())
s := &shards{
qm: t,
queues: queues,
done: make(chan struct{}),
running: int32(numShards),
ctx: ctx,
cancel: cancel,
}
s.wg.Add(numShards)
return s
}
@ -399,11 +408,22 @@ func (s *shards) start() {
}
}
func (s *shards) stop() {
func (s *shards) stop(deadline time.Duration) {
// Attempt a clean shutdown.
for _, shard := range s.queues {
close(shard)
}
s.wg.Wait()
select {
case <-s.done:
return
case <-time.After(deadline):
level.Error(s.qm.logger).Log("msg", "Failed to flush all samples on shutdown")
}
// Force an unclean shutdown.
s.cancel()
<-s.done
return
}
func (s *shards) enqueue(sample *model.Sample) bool {
@ -421,7 +441,12 @@ func (s *shards) enqueue(sample *model.Sample) bool {
}
func (s *shards) runShard(i int) {
defer s.wg.Done()
defer func() {
if atomic.AddInt32(&s.running, -1) == 0 {
close(s.done)
}
}()
queue := s.queues[i]
// Send batches of at most MaxSamplesPerSend samples to the remote storage.
@ -442,6 +467,9 @@ func (s *shards) runShard(i int) {
for {
select {
case <-s.ctx.Done():
return
case sample, ok := <-queue:
if !ok {
if len(pendingSamples) > 0 {
@ -489,7 +517,7 @@ func (s *shards) sendSamplesWithBackoff(samples model.Samples) {
for retries := s.qm.cfg.MaxRetries; retries > 0; retries-- {
begin := time.Now()
req := ToWriteRequest(samples)
err := s.qm.client.Store(req)
err := s.qm.client.Store(s.ctx, req)
sentBatchDuration.WithLabelValues(s.qm.queueName).Observe(time.Since(begin).Seconds())
if err == nil {

View file

@ -14,6 +14,7 @@
package remote
import (
"context"
"fmt"
"reflect"
"sync"
@ -26,6 +27,8 @@ import (
"github.com/prometheus/prometheus/prompb"
)
const defaultFlushDeadline = 1 * time.Minute
type TestStorageClient struct {
receivedSamples map[string][]*prompb.Sample
expectedSamples map[string][]*prompb.Sample
@ -69,7 +72,7 @@ func (c *TestStorageClient) waitForExpectedSamples(t *testing.T) {
}
}
func (c *TestStorageClient) Store(req *prompb.WriteRequest) error {
func (c *TestStorageClient) Store(_ context.Context, req *prompb.WriteRequest) error {
c.mtx.Lock()
defer c.mtx.Unlock()
count := 0
@ -109,7 +112,7 @@ func TestSampleDelivery(t *testing.T) {
cfg := config.DefaultQueueConfig
cfg.MaxShards = 1
m := NewQueueManager(nil, cfg, nil, nil, c)
m := NewQueueManager(nil, cfg, nil, nil, c, defaultFlushDeadline)
// These should be received by the client.
for _, s := range samples[:len(samples)/2] {
@ -145,7 +148,7 @@ func TestSampleDeliveryTimeout(t *testing.T) {
cfg := config.DefaultQueueConfig
cfg.MaxShards = 1
cfg.BatchSendDeadline = 100 * time.Millisecond
m := NewQueueManager(nil, cfg, nil, nil, c)
m := NewQueueManager(nil, cfg, nil, nil, c, defaultFlushDeadline)
m.Start()
defer m.Stop()
@ -181,7 +184,7 @@ func TestSampleDeliveryOrder(t *testing.T) {
c := NewTestStorageClient()
c.expectSamples(samples)
m := NewQueueManager(nil, config.DefaultQueueConfig, nil, nil, c)
m := NewQueueManager(nil, config.DefaultQueueConfig, nil, nil, c, defaultFlushDeadline)
// These should be received by the client.
for _, s := range samples {
@ -209,9 +212,12 @@ func NewTestBlockedStorageClient() *TestBlockingStorageClient {
}
}
func (c *TestBlockingStorageClient) Store(_ *prompb.WriteRequest) error {
func (c *TestBlockingStorageClient) Store(ctx context.Context, _ *prompb.WriteRequest) error {
atomic.AddUint64(&c.numCalls, 1)
<-c.block
select {
case <-c.block:
case <-ctx.Done():
}
return nil
}
@ -259,7 +265,7 @@ func TestSpawnNotMoreThanMaxConcurrentSendsGoroutines(t *testing.T) {
cfg := config.DefaultQueueConfig
cfg.MaxShards = 1
cfg.Capacity = n
m := NewQueueManager(nil, cfg, nil, nil, c)
m := NewQueueManager(nil, cfg, nil, nil, c, defaultFlushDeadline)
m.Start()
@ -299,3 +305,26 @@ func TestSpawnNotMoreThanMaxConcurrentSendsGoroutines(t *testing.T) {
t.Errorf("Saw %d concurrent sends, expected 1", numCalls)
}
}
func TestShutdown(t *testing.T) {
deadline := 10 * time.Second
c := NewTestBlockedStorageClient()
m := NewQueueManager(nil, config.DefaultQueueConfig, nil, nil, c, deadline)
for i := 0; i < config.DefaultQueueConfig.MaxSamplesPerSend; i++ {
m.Append(&model.Sample{
Metric: model.Metric{
model.MetricNameLabel: model.LabelValue(fmt.Sprintf("test_metric_%d", i)),
},
Value: model.SampleValue(i),
Timestamp: model.Time(i),
})
}
m.Start()
start := time.Now()
m.Stop()
duration := time.Now().Sub(start)
if duration > deadline+(deadline/10) {
t.Errorf("Took too long to shutdown: %s > %s", duration, deadline)
}
}

View file

@ -16,6 +16,7 @@ package remote
import (
"context"
"sync"
"time"
"github.com/go-kit/kit/log"
"github.com/prometheus/common/model"
@ -39,14 +40,19 @@ type Storage struct {
// For reads
queryables []storage.Queryable
localStartTimeCallback startTimeCallback
flushDeadline time.Duration
}
// NewStorage returns a remote.Storage.
func NewStorage(l log.Logger, stCallback startTimeCallback) *Storage {
func NewStorage(l log.Logger, stCallback startTimeCallback, flushDeadline time.Duration) *Storage {
if l == nil {
l = log.NewNopLogger()
}
return &Storage{logger: l, localStartTimeCallback: stCallback}
return &Storage{
logger: l,
localStartTimeCallback: stCallback,
flushDeadline: flushDeadline,
}
}
// ApplyConfig updates the state as the new config requires.
@ -74,6 +80,7 @@ func (s *Storage) ApplyConfig(conf *config.Config) error {
conf.GlobalConfig.ExternalLabels,
rwConf.WriteRelabelConfigs,
c,
s.flushDeadline,
))
}