feat(go): improve realtime pipeline resilience

This commit is contained in:
lingniu
2026-07-03 15:20:22 +08:00
parent b8299faefa
commit 9994f2baf4
6 changed files with 213 additions and 11 deletions

View File

@@ -3,6 +3,7 @@ package main
import (
"context"
"encoding/json"
"errors"
"os"
"strings"
"sync"
@@ -103,7 +104,7 @@ func TestAsyncSecondaryRealtimeUpdaterDoesNotBlockPrimaryPath(t *testing.T) {
primary := &contextCheckingRealtimeUpdater{}
secondary := &blockingRealtimeUpdater{started: make(chan struct{}), release: make(chan struct{})}
registry := metrics.NewRegistry()
updater := newAsyncSecondaryRealtimeUpdater(primary, secondary, 8, 1, registry)
updater := newAsyncSecondaryRealtimeUpdater(primary, secondary, 8, 1, registry, nil)
defer updater.Close()
start := time.Now()
@@ -122,8 +123,14 @@ func TestAsyncSecondaryRealtimeUpdaterDoesNotBlockPrimaryPath(t *testing.T) {
t.Fatal("secondary update was not started asynchronously")
}
close(secondary.release)
if text := registry.Render(); !strings.Contains(text, `vehicle_realtime_async_queue_total{protocol="GB32960",status="queued",store="mysql"} 1`) {
t.Fatalf("async queue metric missing:\n%s", text)
text := registry.Render()
for _, want := range []string{
`vehicle_realtime_async_queue_total{protocol="GB32960",status="queued",store="mysql"} 1`,
`vehicle_realtime_async_queue_depth{protocol="GB32960",store="mysql"}`,
} {
if !strings.Contains(text, want) {
t.Fatalf("async queue metric missing %s:\n%s", want, text)
}
}
}
@@ -139,11 +146,49 @@ func TestStoreMetricUpdaterRecordsStoreUpdateResults(t *testing.T) {
if err := updater.Update(context.Background(), env); err != nil {
t.Fatalf("Update() error = %v", err)
}
if text := registry.Render(); !strings.Contains(text, `vehicle_realtime_store_updates_total{protocol="JT808",status="ok",store="redis"} 1`) {
t.Fatalf("store update metric missing:\n%s", text)
text := registry.Render()
for _, want := range []string{
`vehicle_realtime_store_updates_total{protocol="JT808",status="ok",store="redis"} 1`,
`vehicle_realtime_store_update_duration_ms{protocol="JT808",status="ok",store="redis"}`,
} {
if !strings.Contains(text, want) {
t.Fatalf("store update metric missing %s:\n%s", want, text)
}
}
}
func TestRetryRealtimeUpdaterRetriesTransientMySQLDeadlock(t *testing.T) {
delegate := &retryOnceRealtimeUpdater{err: errors.New("Error 1213 (40001): Deadlock found when trying to get lock")}
updater := retryRealtimeUpdater{delegate: delegate, attempts: 3}
if err := updater.Update(context.Background(), envelope.FrameEnvelope{Protocol: envelope.ProtocolGB32960, VIN: "VIN001"}); err != nil {
t.Fatalf("Update() error = %v", err)
}
if delegate.count != 2 {
t.Fatalf("delegate updates = %d, want 2", delegate.count)
}
}
func TestAsyncSecondaryRealtimeUpdaterLogsSecondaryErrors(t *testing.T) {
env := envelope.FrameEnvelope{Protocol: envelope.ProtocolGB32960, VIN: "VIN001", EventID: "event-1"}
logger := &recordingRealtimeLogger{}
updater := newAsyncSecondaryRealtimeUpdater(nil, failingRealtimeUpdater{}, 8, 1, nil, logger)
defer updater.Close()
if err := updater.Update(context.Background(), env); err != nil {
t.Fatalf("Update() error = %v", err)
}
deadline := time.Now().Add(time.Second)
for time.Now().Before(deadline) {
if logger.containsError("async realtime secondary update failed") {
return
}
time.Sleep(10 * time.Millisecond)
}
t.Fatalf("secondary error was not logged: %#v", logger.errors)
}
func TestKafkaTopicsFromEnvDefaultsToGoRawTopics(t *testing.T) {
got := strings.Join(kafkaTopicsFromEnv(), ",")
want := "vehicle.raw.go.gb32960.v1,vehicle.raw.go.jt808.v1,vehicle.raw.go.yutong-mqtt.v1"
@@ -235,6 +280,30 @@ func (discardRealtimeLogger) Info(string, ...any) {}
func (discardRealtimeLogger) Error(string, ...any) {}
func (discardRealtimeLogger) Warn(string, ...any) {}
type recordingRealtimeLogger struct {
mu sync.Mutex
errors []string
}
func (l *recordingRealtimeLogger) Info(string, ...any) {}
func (l *recordingRealtimeLogger) Warn(string, ...any) {}
func (l *recordingRealtimeLogger) Error(message string, _ ...any) {
l.mu.Lock()
defer l.mu.Unlock()
l.errors = append(l.errors, message)
}
func (l *recordingRealtimeLogger) containsError(message string) bool {
l.mu.Lock()
defer l.mu.Unlock()
for _, got := range l.errors {
if got == message {
return true
}
}
return false
}
type blockingRealtimeUpdater struct {
started chan struct{}
release chan struct{}
@@ -246,3 +315,24 @@ func (u *blockingRealtimeUpdater) Update(context.Context, envelope.FrameEnvelope
<-u.release
return nil
}
type failingRealtimeUpdater struct{}
func (failingRealtimeUpdater) Update(context.Context, envelope.FrameEnvelope) error {
return errTestRealtimeUpdate
}
var errTestRealtimeUpdate = errors.New("test realtime update failed")
type retryOnceRealtimeUpdater struct {
err error
count int
}
func (u *retryOnceRealtimeUpdater) Update(context.Context, envelope.FrameEnvelope) error {
u.count++
if u.count == 1 {
return u.err
}
return nil
}