Zhen Ye c7b5c23ff6
enhance: filter the empty timetick from consuming side (#46541)
issue: #46540

Empty timetick is just used to sync up the time clock between different
component in milvus. So empty timetick can be ignored if we achieve the
lsn/mvcc semantic for timetick. Currently, some components need the
empty timetick to trigger some operation, such as flush/tsafe. So we
only slow down the empty time tick for 5 seconds.

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
- Core invariant: with LSN/MVCC semantics consumers only need (a) the
first timetick that advances the latest-required-MVCC to unblock
MVCC-dependent waits and (b) occasional periodic timeticks (~≤5s) for
clock synchronization—therefore frequent non-persisted empty timeticks
can be suppressed without breaking MVCC correctness.
- Logic removed/simplified: per-message dispatch/consumption of frequent
non-persisted empty timeticks is suppressed — an MVCC-aware filter
emptyTimeTickSlowdowner (internal/util/pipeline/consuming_slowdown.go)
short-circuits frequent empty timeticks in the stream pipeline
(internal/util/pipeline/stream_pipeline.go), and the WAL flusher
rate-limits non-persisted timetick dispatch to one emission per ~5s
(internal/streamingnode/server/flusher/flusherimpl/wal_flusher.go); the
delegator exposes GetLatestRequiredMVCCTimeTick to drive the filter
(internal/querynodev2/delegator/delegator.go).
- Why this does NOT introduce data loss or regressions: the slowdowner
always refreshes latestRequiredMVCCTimeTick via
GetLatestRequiredMVCCTimeTick and (1) never filters timeticks <
latestRequiredMVCCTimeTick (so existing tsafe/flush waits stay
unblocked) and (2) always lets the first timetick ≥
latestRequiredMVCCTimeTick pass to notify pending MVCC waits;
separately, WAL flusher suppression applies only to non-persisted
timeticks and still emits when the 5s threshold elapses, preserving
periodic clock-sync messages used by flush/tsafe.
- Enhancement summary (where it takes effect): adds
GetLatestRequiredMVCCTimeTick on ShardDelegator and
LastestMVCCTimeTickGetter, wires emptyTimeTickSlowdowner into
NewPipelineWithStream (internal/util/pipeline), and adds WAL flusher
rate-limiting + metrics
(internal/streamingnode/server/flusher/flusherimpl/wal_flusher.go,
pkg/metrics) to reduce CPU/dispatch overhead while keeping MVCC
correctness and periodic synchronization.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

---------

Signed-off-by: chyezh <chyezh@outlook.com>
2026-01-06 20:53:24 +08:00

126 lines
3.4 KiB
Go

package helper
import (
"context"
"flag"
"testing"
"time"
"go.uber.org/zap"
"google.golang.org/grpc"
client "github.com/milvus-io/milvus/client/v2/milvusclient"
"github.com/milvus-io/milvus/pkg/v2/log"
"github.com/milvus-io/milvus/tests/go_client/base"
"github.com/milvus-io/milvus/tests/go_client/common"
)
var (
addr = flag.String("addr", "http://localhost:19530", "server host and port")
user = flag.String("user", "root", "user")
password = flag.String("password", "Milvus", "password")
logLevel = flag.String("log.level", "info", "log level for test")
teiEndpoint = flag.String("tei_endpoint", "http://text-embeddings-service.milvus-ci.svc.cluster.local:80", "TEI service endpoint for text embedding tests")
teiRerankerEndpoint = flag.String("tei_reranker_uri", "http://text-rerank-service.milvus-ci.svc.cluster.local:80", "TEI reranker service endpoint")
teiModelDim = flag.Int("tei_model_dim", 768, "Vector dimension for text embedding model")
defaultClientConfig *client.ClientConfig
)
func setDefaultClientConfig(cfg *client.ClientConfig) {
defaultClientConfig = cfg
}
func GetDefaultClientConfig() *client.ClientConfig {
newCfg := *defaultClientConfig
dialOptions := newCfg.DialOptions
newDialOptions := make([]grpc.DialOption, len(dialOptions))
copy(newDialOptions, dialOptions)
newCfg.DialOptions = newDialOptions
return &newCfg
}
func GetAddr() string {
return *addr
}
func GetUser() string {
return *user
}
func GetPassword() string {
return *password
}
func GetTEIEndpoint() string {
return *teiEndpoint
}
func GetTEIRerankerEndpoint() string {
return *teiRerankerEndpoint
}
func GetTEIModelDim() int {
return *teiModelDim
}
func parseLogConfig() {
log.Info("Parser Log Level", zap.String("logLevel", *logLevel))
switch *logLevel {
case "debug", "DEBUG", "Debug":
log.SetLevel(zap.DebugLevel)
case "info", "INFO", "Info":
log.SetLevel(zap.InfoLevel)
case "warn", "WARN", "Warn":
log.SetLevel(zap.WarnLevel)
case "error", "ERROR", "Error":
log.SetLevel(zap.ErrorLevel)
default:
log.SetLevel(zap.InfoLevel)
}
}
func setup() {
log.Info("Start to setup all......")
flag.Parse()
parseLogConfig()
log.Info("Parser Milvus address", zap.String("address", *addr))
// set default milvus client config
setDefaultClientConfig(&client.ClientConfig{Address: *addr})
}
// Teardown teardown
func teardown() {
log.Info("Start to tear down all.....")
ctx, cancel := context.WithTimeout(context.Background(), time.Second*common.DefaultTimeout)
defer cancel()
mc, err := base.NewMilvusClient(ctx, &client.ClientConfig{Address: GetAddr(), Username: GetUser(), Password: GetPassword()})
if err != nil {
log.Error("teardown failed to connect milvus with error", zap.Error(err))
}
defer mc.Close(ctx)
// clear dbs
dbs, _ := mc.ListDatabase(ctx, client.NewListDatabaseOption())
for _, db := range dbs {
if db != common.DefaultDb {
_ = mc.UseDatabase(ctx, client.NewUseDatabaseOption(db))
collections, _ := mc.ListCollections(ctx, client.NewListCollectionOption())
for _, coll := range collections {
_ = mc.DropCollection(ctx, client.NewDropCollectionOption(coll))
}
_ = mc.DropDatabase(ctx, client.NewDropDatabaseOption(db))
}
}
}
func RunTests(m *testing.M) int {
setup()
code := m.Run()
if code != 0 {
log.Error("Tests failed and exited", zap.Int("code", code))
}
teardown()
return code
}