Skip to content

kafka: fix metrics #5873

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
3AceShowHand wants to merge 1 commit into
base: master
Choose a base branch
from
Open

kafka: fix metrics #5873

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 1 addition & 14 deletions downstreamadapter/sink/cloudstorage/sink.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -328,17 +328,7 @@ func (s *sink) AddCheckpointTs(ts uint64) {
}

func (s *sink) sendCheckpointTs(ctx context.Context) error {
var (
keyspace = s.changefeedID.Keyspace()
changefeed = s.changefeedID.Name()
)
checkpointTsMessageDuration := metrics.CheckpointTsMessageDuration.WithLabelValues(keyspace, changefeed)
checkpointTsMessageCount := metrics.CheckpointTsMessageCount.WithLabelValues(keyspace, changefeed)
defer func() {
metrics.CheckpointTsMessageDuration.DeleteLabelValues(keyspace, changefeed)
metrics.CheckpointTsMessageCount.DeleteLabelValues(keyspace, changefeed)
}()

keyspace, changefeed := s.changefeedID.Keyspace(), s.changefeedID.Name()
var checkpoint uint64
for {
select {
Expand Down Expand Up @@ -376,9 +366,6 @@ func (s *sink) sendCheckpointTs(ctx context.Context) error {
}
s.lastSendCheckpointTsTime = time.Now()
s.lastCheckpointTs.Store(checkpoint)

checkpointTsMessageCount.Inc()
checkpointTsMessageDuration.Observe(time.Since(start).Seconds())
}
}

Expand Down
10 changes: 0 additions & 10 deletions downstreamadapter/sink/kafka/sink.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -503,13 +503,6 @@ func (s *sink) AddCheckpointTs(ts uint64) {
}

func (s *sink) sendCheckpoint(ctx context.Context) error {
checkpointTsMessageDuration := metrics.CheckpointTsMessageDuration.WithLabelValues(s.changefeedID.Keyspace(), s.changefeedID.Name())
checkpointTsMessageCount := metrics.CheckpointTsMessageCount.WithLabelValues(s.changefeedID.Keyspace(), s.changefeedID.Name())
defer func() {
metrics.CheckpointTsMessageDuration.DeleteLabelValues(s.changefeedID.Keyspace(), s.changefeedID.Name())
metrics.CheckpointTsMessageCount.DeleteLabelValues(s.changefeedID.Keyspace(), s.changefeedID.Name())
}()

var (
msg *codecCommon.Message
partitionNum int32
Expand All @@ -524,7 +517,6 @@ func (s *sink) sendCheckpoint(ctx context.Context) error {
return nil
}

start := time.Now()
msg, err = s.comp.encoder.EncodeCheckpointEvent(ts)
if err != nil {
return err
Expand Down Expand Up @@ -561,8 +553,6 @@ func (s *sink) sendCheckpoint(ctx context.Context) error {
}
}
}
checkpointTsMessageCount.Inc()
checkpointTsMessageDuration.Observe(time.Since(start).Seconds())
}
}
}
Expand Down
52 changes: 5 additions & 47 deletions downstreamadapter/sink/metrics/mq.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,65 +16,25 @@ package metrics
import (
"sync"

"github.com/pingcap/ticdc/pkg/sink/codec"
"github.com/pingcap/ticdc/pkg/sink/kafka"
"github.com/pingcap/ticdc/pkg/sink/kafka/claimcheck"
"github.com/prometheus/client_golang/prometheus"
)

var (
// WorkerSendMessageDuration records the duration of flushing a group messages.
WorkerSendMessageDuration = prometheus.NewHistogramVec(
prometheus.HistogramOpts{
Namespace: "ticdc",
Subsystem: "sink",
Name: "mq_worker_send_message_duration",
Help: "Send Message duration(s) for MQ worker.",
Buckets: prometheus.ExponentialBuckets(0.001, 2, 20), // 1ms~524s
}, []string{"namespace", "changefeed"})
// WorkerBatchSize record the size of each batched messages.
WorkerBatchSize = prometheus.NewHistogramVec(
prometheus.HistogramOpts{
Namespace: "ticdc",
Subsystem: "sink",
Name: "mq_worker_batch_size",
Help: "Batch size for MQ worker.",
Buckets: prometheus.ExponentialBuckets(4, 2, 10), // 4 ~ 2048
}, []string{"namespace", "changefeed"})
// WorkerBatchDuration record the time duration cost on batch messages.
WorkerBatchDuration = prometheus.NewHistogramVec(
prometheus.HistogramOpts{
Namespace: "ticdc",
Subsystem: "sink",
Name: "mq_worker_batch_duration",
Help: "Batch duration for MQ worker.",
Buckets: prometheus.ExponentialBuckets(0.004, 2, 10), // 4ms ~ 2s
}, []string{"namespace", "changefeed"})
)

var (
mqServerRegistryMu sync.RWMutex
// mqServerRegistry is shared by all MQ sinks on the node. Bootstrap can now
// create multiple changefeeds concurrently, so both reads and the fallback
// initialization must be synchronized to avoid racing on the global pointer.
// mqServerRegistry is shared by all MQ sinks on the node. Bootstrap can
// create multiple changefeeds concurrently, so reads and initialization
// must be synchronized.
mqServerRegistry *prometheus.Registry
)

// InitMQMetrics registers all metrics in this file.
// InitMQMetrics configures the registry used by MQ client metrics.
func InitMQMetrics(registry *prometheus.Registry) {
mqServerRegistryMu.Lock()
mqServerRegistry = registry
mqServerRegistryMu.Unlock()

registry.MustRegister(WorkerSendMessageDuration)
registry.MustRegister(WorkerBatchSize)
registry.MustRegister(WorkerBatchDuration)
claimcheck.InitMetrics(registry)
codec.InitMetrics(registry)
kafka.InitMetrics(registry)
}

// GetMQMetricRegistry for add pulsar default metrics
// GetMQMetricRegistry returns the registry used by MQ client metrics.
func GetMQMetricRegistry() *prometheus.Registry {
mqServerRegistryMu.RLock()
registry := mqServerRegistry
Expand All @@ -85,8 +45,6 @@ func GetMQMetricRegistry() *prometheus.Registry {

mqServerRegistryMu.Lock()
defer mqServerRegistryMu.Unlock()
// Make sure registry is not nil when MQ sink metrics are first requested
// before the server metrics bootstrap wires in the shared registry.
if mqServerRegistry == nil {
mqServerRegistry = prometheus.DefaultRegisterer.(*prometheus.Registry)
}
Expand Down
11 changes: 0 additions & 11 deletions downstreamadapter/sink/pulsar/sink.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -283,13 +283,6 @@ func (s *sink) SetTableSchemaStore(tableSchemaStore *commonEvent.TableSchemaStor
}

func (s *sink) sendCheckpoint(ctx context.Context) error {
checkpointTsMessageDuration := metrics.CheckpointTsMessageDuration.WithLabelValues(s.changefeedID.Keyspace(), s.changefeedID.Name())
checkpointTsMessageCount := metrics.CheckpointTsMessageCount.WithLabelValues(s.changefeedID.Keyspace(), s.changefeedID.Name())

defer func() {
metrics.CheckpointTsMessageDuration.DeleteLabelValues(s.changefeedID.Keyspace(), s.changefeedID.Name())
metrics.CheckpointTsMessageCount.DeleteLabelValues(s.changefeedID.Keyspace(), s.changefeedID.Name())
}()
var (
msg *common.Message
err error
Expand All @@ -306,7 +299,6 @@ func (s *sink) sendCheckpoint(ctx context.Context) error {
return nil
}

start := time.Now()
msg, err = s.comp.encoder.EncodeCheckpointEvent(ts)
if err != nil {
return errors.Trace(err)
Expand Down Expand Up @@ -344,9 +336,6 @@ func (s *sink) sendCheckpoint(ctx context.Context) error {
}
}
}

checkpointTsMessageCount.Inc()
checkpointTsMessageDuration.Observe(time.Since(start).Seconds())
}
}
}
Expand Down
19 changes: 0 additions & 19 deletions pkg/metrics/sink.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -144,23 +144,6 @@ var (
Help: "Batch duration for MQ worker.",
Buckets: prometheus.ExponentialBuckets(0.004, 2, 10), // 4ms ~ 2s
}, []string{GetKeyspaceLabel(), "changefeed"})

CheckpointTsMessageDuration = prometheus.NewHistogramVec(
prometheus.HistogramOpts{
Namespace: "ticdc",
Subsystem: "sink",
Name: "mq_checkpoint_ts_message_duration",
Help: "Duration of sending checkpoint ts message.",
Buckets: prometheus.ExponentialBuckets(0.001, 2, 20), // 1ms~524s
}, []string{GetKeyspaceLabel(), "changefeed"})

CheckpointTsMessageCount = prometheus.NewGaugeVec(
prometheus.GaugeOpts{
Namespace: "ticdc",
Subsystem: "sink",
Name: "mq_checkpoint_ts_message_count",
Help: "Number of checkpoint ts messages sent.",
}, []string{GetKeyspaceLabel(), "changefeed"})
)

// InitMetrics registers all metrics in this file.
Expand All @@ -183,8 +166,6 @@ func initSinkMetrics(registry *prometheus.Registry) {
registry.MustRegister(WorkerSendMessageDuration)
registry.MustRegister(WorkerBatchSize)
registry.MustRegister(WorkerBatchDuration)
registry.MustRegister(CheckpointTsMessageDuration)
registry.MustRegister(CheckpointTsMessageCount)

// pulsar sink metrics
initPulsarMetrics(registry)
Expand Down
Loading