diff --git a/docs/ops/go-vehicle-ingest-memory.md b/docs/ops/go-vehicle-ingest-memory.md index d46813e3..1d7e1558 100644 --- a/docs/ops/go-vehicle-ingest-memory.md +++ b/docs/ops/go-vehicle-ingest-memory.md @@ -440,7 +440,7 @@ FAST_WRITER_TDENGINE_MAX_OPEN_CONNS FAST_WRITER_TDENGINE_MAX_IDLE_CONNS ``` -默认 `max_open=1`、`max_idle=1`。原因是当前 TDengine 写入依赖启动时执行的 `USE `;`database/sql` 连接池新建连接不会继承其他连接上的 `USE` 状态。实测默认跟随 worker 放大到 8 会导致 `[0x200] db is not specified` / `[0x2616] Database not specified`。后续只有在 TDengine DSN 或 writer SQL 改为每条连接都明确选库后,才能提高该连接池。 +默认 `max_open=1`、`max_idle=1`,生产可以通过 env 小步放大。此前实测默认跟随 worker 放大到 8 会导致 `[0x200] db is not specified` / `[0x2616] Database not specified`,根因是 TDengine `USE ` 不会自动作用到 `database/sql` 新建连接。2026-07-03 已将 fast-writer 使用的 history writer 改为生成 `database.table` 形式的 database-qualified SQL,并在 ECS 上将 pool 调到 `2/2` 验证通过:未再出现选库错误,写入 ok 计数持续增长。 同时新增 batch pending gauge: diff --git a/docs/ops/vehicle-ingest-runbook.md b/docs/ops/vehicle-ingest-runbook.md index 94d617d1..93b57ea5 100644 --- a/docs/ops/vehicle-ingest-runbook.md +++ b/docs/ops/vehicle-ingest-runbook.md @@ -136,7 +136,7 @@ go run ./cmd/load-sim \ | `vehicle_fast_writer_nats_consumer_pending` | 持续增长且 `> 10000` | fast-writer 消费 NATS 的速度跟不上入口写入速度。 | | `vehicle_fast_writer_batch_pending_messages` | 持续非 0 或 burst 后不回落 | fast-writer 已拉取 NATS 消息但尚未完成 TDengine/Redis 写入和 ack。 | | `vehicle_fast_writer_batch_pending_envelopes` | 持续非 0 或 burst 后不回落 | fast-writer 当前批次已有有效 envelope 在等待落库或 ack。 | -| `vehicle_fast_writer_stage_duration_ms_histogram_bucket` | 某个 stage 的 p99 连续 5 分钟上升 | NATS 快速写链路在 TDengine、Redis 或 NATS ack 某一阶段变慢;当前 TDengine 连接池默认保持 `1`,放大 `FAST_WRITER_TDENGINE_MAX_OPEN_CONNS` 前必须先确认每条连接都能明确选库。 | +| `vehicle_fast_writer_stage_duration_ms_histogram_bucket` | 某个 stage 的 p99 连续 5 分钟上升 | NATS 快速写链路在 TDengine、Redis 或 NATS ack 某一阶段变慢;TDengine 阶段可小步调整 `FAST_WRITER_TDENGINE_MAX_OPEN_CONNS`,调整后必须观察是否出现选库错误和 ack-pending 增长。 | | `vehicle_realtime_store_update_duration_ms_histogram_bucket{store="redis"}` | p99 连续 5 分钟上升 | Redis 实时投影变慢,会直接影响 realtime consumer 追平能力。 | | `vehicle_realtime_store_update_duration_ms_histogram_bucket{store="mysql"}` | p99 连续 5 分钟上升 | MySQL 当前态/位置投影变慢,需结合 async queue depth 和 dropped 计数判断。 | | `vehicle_stat_write_duration_ms_histogram_bucket` | p99 连续 5 分钟上升 | MySQL 每日里程统计写入变慢,可能导致 stat Kafka lag 增长。 | diff --git a/go/vehicle-gateway/cmd/nats-fast-writer/main.go b/go/vehicle-gateway/cmd/nats-fast-writer/main.go index e7fc1013..2013d9c7 100644 --- a/go/vehicle-gateway/cmd/nats-fast-writer/main.go +++ b/go/vehicle-gateway/cmd/nats-fast-writer/main.go @@ -72,19 +72,13 @@ func main() { logger.Error("tdengine ping failed", "error", err) os.Exit(1) } - historyWriter := history.NewWriter(tdDB) + historyWriter := history.NewWriterWithDatabase(tdDB, cfg.TDengineDatabase) if cfg.TDengineEnsureSchema { if err := historyWriter.EnsureSchema(ctx, cfg.TDengineDatabase); err != nil { logger.Error("tdengine schema bootstrap failed", "error", err) os.Exit(1) } } - if strings.TrimSpace(cfg.TDengineDatabase) != "" { - if _, err := tdDB.ExecContext(ctx, "USE "+cfg.TDengineDatabase); err != nil { - logger.Error("tdengine use database failed", "database", cfg.TDengineDatabase, "error", err) - os.Exit(1) - } - } redisClient := redis.NewClient(&redis.Options{ Addr: cfg.RedisAddr, diff --git a/go/vehicle-gateway/internal/history/writer.go b/go/vehicle-gateway/internal/history/writer.go index 69bce4fb..7e627c47 100644 --- a/go/vehicle-gateway/internal/history/writer.go +++ b/go/vehicle-gateway/internal/history/writer.go @@ -22,8 +22,9 @@ type Execer interface { } type Writer struct { - exec Execer - cache tableCache + exec Execer + database string + cache tableCache } const ( @@ -50,6 +51,12 @@ func NewWriter(exec Execer) *Writer { return &Writer{exec: exec, cache: tableCache{seen: map[string]struct{}{}}} } +func NewWriterWithDatabase(exec Execer, database string) *Writer { + writer := NewWriter(exec) + writer.database = normalizeIdentifier(database) + return writer +} + func (w *Writer) EnsureSchema(ctx context.Context, database string) error { for _, statement := range SchemaStatements(database) { if _, err := w.exec.ExecContext(ctx, statement); err != nil { @@ -59,6 +66,14 @@ func (w *Writer) EnsureSchema(ctx context.Context, database string) error { return nil } +func (w *Writer) qualify(table string) string { + table = normalizeIdentifier(table) + if w.database == "" { + return table + } + return w.database + "." + table +} + func (w *Writer) AppendAll(ctx context.Context, env envelope.FrameEnvelope) error { if err := w.AppendRawFrame(ctx, env); err != nil { return err @@ -87,7 +102,7 @@ func (w *Writer) AppendRawFrame(ctx context.Context, env envelope.FrameEnvelope) _, err := w.exec.ExecContext(ctx, fmt.Sprintf(`INSERT INTO %s (ts, frame_id, event_id, message_id, event_time, received_at, raw_size_bytes, raw_hex, raw_text, parsed_json, parse_status, parse_error, source_endpoint) -VALUES (%s)`, table, joinLiterals(rawValues(env, rawHex, rawText, parsedFields)))) +VALUES (%s)`, w.qualify(table), joinLiterals(rawValues(env, rawHex, rawText, parsedFields)))) if err != nil { return err } @@ -103,7 +118,7 @@ VALUES (%s)`, table, joinLiterals(rawValues(env, rawHex, rawText, parsedFields)) for _, chunk := range chunks { if _, err := w.exec.ExecContext(ctx, fmt.Sprintf(`INSERT INTO %s (ts, event_id, frame_id, received_at, payload_kind, chunk_index, chunk_count, chunk_text) -VALUES (%s)`, chunkTable, joinLiterals(chunkValues(env, chunk)))); err != nil { +VALUES (%s)`, w.qualify(chunkTable), joinLiterals(chunkValues(env, chunk)))); err != nil { return err } } @@ -144,7 +159,7 @@ func (w *Writer) AppendRawFrameBatch(ctx context.Context, envelopes []envelope.F if _, err := w.exec.ExecContext(ctx, fmt.Sprintf(`INSERT INTO %s (ts, frame_id, event_id, message_id, event_time, received_at, raw_size_bytes, raw_hex, raw_text, parsed_json, parse_status, parse_error, source_endpoint) -VALUES %s`, table, strings.Join(rows, ","))); err != nil { +VALUES %s`, w.qualify(table), strings.Join(rows, ","))); err != nil { return err } } @@ -155,7 +170,7 @@ VALUES %s`, table, strings.Join(rows, ","))); err != nil { _ = chunkEnvByTable[table] if _, err := w.exec.ExecContext(ctx, fmt.Sprintf(`INSERT INTO %s (ts, event_id, frame_id, received_at, payload_kind, chunk_index, chunk_count, chunk_text) -VALUES %s`, table, strings.Join(rows, ","))); err != nil { +VALUES %s`, w.qualify(table), strings.Join(rows, ","))); err != nil { return err } } @@ -178,7 +193,7 @@ func (w *Writer) AppendLocation(ctx context.Context, env envelope.FrameEnvelope) _, err := w.exec.ExecContext(ctx, fmt.Sprintf(`INSERT INTO %s (ts, event_id, received_at, longitude, latitude, altitude_m, speed_kmh, direction_deg, alarm_flag, status_flag, total_mileage_km) -VALUES (%s)`, table, joinLiterals(locationValues(env, longitude, latitude)))) +VALUES (%s)`, w.qualify(table), joinLiterals(locationValues(env, longitude, latitude)))) return err } @@ -206,7 +221,7 @@ func (w *Writer) AppendLocationBatch(ctx context.Context, envelopes []envelope.F if _, err := w.exec.ExecContext(ctx, fmt.Sprintf(`INSERT INTO %s (ts, event_id, received_at, longitude, latitude, altitude_m, speed_kmh, direction_deg, alarm_flag, status_flag, total_mileage_km) -VALUES %s`, table, strings.Join(rows, ","))); err != nil { +VALUES %s`, w.qualify(table), strings.Join(rows, ","))); err != nil { return err } } @@ -222,7 +237,7 @@ func (w *Writer) ensureRawChild(ctx context.Context, table string, stable string return nil } statement := fmt.Sprintf("CREATE TABLE IF NOT EXISTS %s USING %s TAGS ('%s', '%s', '%s', '%s', '%s')", - table, stable, + w.qualify(table), w.qualify(stable), quote(string(env.Protocol)), quote(env.VehicleKey()), quote(env.VIN), @@ -232,7 +247,7 @@ func (w *Writer) ensureRawChild(ctx context.Context, table string, stable string return err } if phone := normalizePhoneTag(env.Phone); phone != "" { - if _, err := w.exec.ExecContext(ctx, fmt.Sprintf("ALTER TABLE %s SET TAG phone = '%s'", table, quote(phone))); err != nil { + if _, err := w.exec.ExecContext(ctx, fmt.Sprintf("ALTER TABLE %s SET TAG phone = '%s'", w.qualify(table), quote(phone))); err != nil { return err } } @@ -250,8 +265,9 @@ func (w *Writer) ensureLocationChild(ctx context.Context, table string, env enve if ok { return nil } - statement := fmt.Sprintf("CREATE TABLE IF NOT EXISTS %s USING vehicle_locations TAGS ('%s', '%s')", - table, + statement := fmt.Sprintf("CREATE TABLE IF NOT EXISTS %s USING %s TAGS ('%s', '%s')", + w.qualify(table), + w.qualify("vehicle_locations"), quote(string(env.Protocol)), quote(strings.TrimSpace(env.VIN))) if _, err := w.exec.ExecContext(ctx, statement); err != nil { @@ -367,6 +383,12 @@ func locationTableName(env envelope.FrameEnvelope) string { return "loc_" + strings.ToLower(string(env.Protocol)) + "_" + hash16(strings.TrimSpace(env.VIN)) } +func normalizeIdentifier(value string) string { + value = strings.TrimSpace(value) + value = strings.Trim(value, "`") + return value +} + func normalizePhoneTag(phone string) string { trimmed := strings.TrimLeft(strings.TrimSpace(phone), "0") if trimmed == "" { diff --git a/go/vehicle-gateway/internal/history/writer_test.go b/go/vehicle-gateway/internal/history/writer_test.go index 150f819c..7e8a6e8f 100644 --- a/go/vehicle-gateway/internal/history/writer_test.go +++ b/go/vehicle-gateway/internal/history/writer_test.go @@ -261,6 +261,30 @@ func TestWriterAppendsBatchRowsByChildTable(t *testing.T) { } } +func TestWriterWithDatabaseQualifiesTDengineTables(t *testing.T) { + exec := &recordingExec{} + writer := NewWriterWithDatabase(exec, "vehicle_ts") + env := sampleEnvelope() + + if err := writer.AppendAll(context.Background(), env); err != nil { + t.Fatalf("AppendAll() error = %v", err) + } + + for _, want := range []string{ + "CREATE TABLE IF NOT EXISTS vehicle_ts.raw_", + "USING vehicle_ts.raw_frames", + "ALTER TABLE vehicle_ts.raw_", + "INSERT INTO vehicle_ts.raw_", + "CREATE TABLE IF NOT EXISTS vehicle_ts.loc_", + "USING vehicle_ts.vehicle_locations", + "INSERT INTO vehicle_ts.loc_", + } { + if !containsSQL(exec.calls, want) { + t.Fatalf("qualified sql missing %q, calls=%v", want, exec.calls) + } + } +} + func TestWriterAppendAllBatchSkipsEmptyBatch(t *testing.T) { exec := &recordingExec{} writer := NewWriter(exec) @@ -338,6 +362,10 @@ func findSQL(calls []execCall, pattern string) string { return "" } +func containsSQL(calls []execCall, pattern string) bool { + return findSQL(calls, pattern) != "" +} + func between(value string, start string, end string) string { startIndex := strings.Index(value, start) if startIndex < 0 {