refactor(server): standarderlize metrics and trace with OTEL (#5054)

you can now export span to Zipkin and metrics to Prometheus when developing locally
follow the docs of OTEL: https://opentelemetry.io/docs/instrumentation/js/exporters/

<img width="2357" alt="image" src="https://github.com/toeverything/AFFiNE/assets/8281226/ec615e1f-3e91-43f7-9111-d7d2629e9679">

.github/helm/affine/charts/graphql/templates/monitoring.yaml

@@ -1,13 +0,0 @@
-{{- if .Values.global.gke.enabled -}}
-apiVersion: monitoring.googleapis.com/v1
-kind: PodMonitoring
-metadata:
-  name: "{{ .Chart.Name }}-monitoring"
-spec:
-  selector:
-    matchLabels:
-      app.kubernetes.io/name: "{{ include "graphql.name" . }}"
-  endpoints:
-    - port: {{ .Values.service.port }}
-      interval: 30s
-{{- end }}

.github/helm/affine/charts/sync/templates/monitoring.yaml

@@ -1,13 +0,0 @@
-{{- if .Values.global.gke.enabled -}}
-apiVersion: monitoring.googleapis.com/v1
-kind: PodMonitoring
-metadata:
-  name: "{{ .Chart.Name }}-monitoring"
-spec:
-  selector:
-    matchLabels:
-      app.kubernetes.io/name: "{{ include "sync.name" . }}"
-  endpoints:
-    - port: {{ .Values.service.port }}
-      interval: 30s
-{{- end }}

packages/backend/server/package.json

@@ -39,6 +39,9 @@
     "@node-rs/jsonwebtoken": "^0.2.3",
     "@opentelemetry/api": "^1.7.0",
     "@opentelemetry/core": "^1.18.1",
+    "@opentelemetry/exporter-prometheus": "^0.45.1",
+    "@opentelemetry/exporter-zipkin": "^1.18.1",
+    "@opentelemetry/host-metrics": "^0.33.2",
     "@opentelemetry/instrumentation": "^0.45.1",
     "@opentelemetry/instrumentation-graphql": "^0.36.0",
     "@opentelemetry/instrumentation-http": "^0.45.1",

packages/backend/server/src/app.ts

@@ -3,7 +3,6 @@ import { Module } from '@nestjs/common';
 import { AppController } from './app.controller';
 import { CacheModule } from './cache';
 import { ConfigModule } from './config';
-import { MetricsModule } from './metrics';
 import { BusinessModules } from './modules';
 import { AuthModule } from './modules/auth';
 import { PrismaModule } from './prisma';
@@ -16,7 +15,6 @@ const BasicModules = [
   ConfigModule.forRoot(),
   CacheModule,
   StorageModule.forRoot(),
-  MetricsModule,
   SessionModule,
   RateLimiterModule,
   AuthModule,

packages/backend/server/src/graphql.module.ts

@@ -8,14 +8,13 @@ import { fileURLToPath } from 'url';
 
 import { Config } from './config';
 import { GQLLoggerPlugin } from './graphql/logger-plugin';
-import { Metrics } from './metrics/metrics';
 
 @Global()
 @Module({
   imports: [
     GraphQLModule.forRootAsync<ApolloDriverConfig>({
       driver: ApolloDriver,
-      useFactory: (config: Config, metrics: Metrics) => {
+      useFactory: (config: Config) => {
         return {
           ...config.graphql,
           path: `${config.path}/graphql`,
@@ -31,10 +30,10 @@ import { Metrics } from './metrics/metrics';
             req,
             res,
           }),
-          plugins: [new GQLLoggerPlugin(metrics)],
+          plugins: [new GQLLoggerPlugin()],
         };
       },
-      inject: [Config, Metrics],
+      inject: [Config],
     }),
   ],
 })

packages/backend/server/src/graphql/logger-plugin.ts

@@ -7,40 +7,39 @@ import { Plugin } from '@nestjs/apollo';
 import { Logger } from '@nestjs/common';
 import { Response } from 'express';
 
-import { Metrics } from '../metrics/metrics';
+import { metrics } from '../metrics/metrics';
 import { ReqContext } from '../types';
 
 @Plugin()
 export class GQLLoggerPlugin implements ApolloServerPlugin {
   protected logger = new Logger(GQLLoggerPlugin.name);
 
-  constructor(private readonly metrics: Metrics) {}
-
   requestDidStart(
     reqContext: GraphQLRequestContext<ReqContext>
   ): Promise<GraphQLRequestListener<GraphQLRequestContext<ReqContext>>> {
     const res = reqContext.contextValue.req.res as Response;
     const operation = reqContext.request.operationName;
 
-    this.metrics.gqlRequest(1, { operation });
+    metrics().gqlRequest.add(1, { operation });
-    const timer = this.metrics.gqlTimer({ operation });
+    const start = Date.now();
 
     return Promise.resolve({
       willSendResponse: () => {
-        const costInMilliseconds = timer() * 1000;
+        const costInMilliseconds = Date.now() - start;
         res.setHeader(
           'Server-Timing',
           `gql;dur=${costInMilliseconds};desc="GraphQL"`
         );
+        metrics().gqlTimer.record(costInMilliseconds, { operation });
         return Promise.resolve();
       },
       didEncounterErrors: () => {
-        this.metrics.gqlError(1, { operation });
+        const costInMilliseconds = Date.now() - start;
-        const costInMilliseconds = timer() * 1000;
         res.setHeader(
           'Server-Timing',
           `gql;dur=${costInMilliseconds};desc="GraphQL ${operation}"`
         );
+        metrics().gqlTimer.record(costInMilliseconds, { operation });
         return Promise.resolve();
       },
     });

packages/backend/server/src/index.ts

@@ -1,22 +1,9 @@
 /// <reference types="./global.d.ts" />
-import { MetricExporter } from '@google-cloud/opentelemetry-cloud-monitoring-exporter';
+import { start as startAutoMetrics } from './metrics';
-import { TraceExporter } from '@google-cloud/opentelemetry-cloud-trace-exporter';
+startAutoMetrics();
+
 import { NestFactory } from '@nestjs/core';
 import type { NestExpressApplication } from '@nestjs/platform-express';
-import {
-  CompositePropagator,
-  W3CBaggagePropagator,
-  W3CTraceContextPropagator,
-} from '@opentelemetry/core';
-import gql from '@opentelemetry/instrumentation-graphql';
-import { HttpInstrumentation } from '@opentelemetry/instrumentation-http';
-import ioredis from '@opentelemetry/instrumentation-ioredis';
-import { NestInstrumentation } from '@opentelemetry/instrumentation-nestjs-core';
-import socketIO from '@opentelemetry/instrumentation-socket.io';
-import { PeriodicExportingMetricReader } from '@opentelemetry/sdk-metrics';
-import { NodeSDK } from '@opentelemetry/sdk-node';
-import { BatchSpanProcessor } from '@opentelemetry/sdk-trace-node';
-import { PrismaInstrumentation } from '@prisma/instrumentation';
 import cookieParser from 'cookie-parser';
 import { static as staticMiddleware } from 'express';
 import graphqlUploadExpress from 'graphql-upload/graphqlUploadExpress.mjs';
@@ -28,35 +15,6 @@ import { serverTimingAndCache } from './middleware/timing';
 import { RedisIoAdapter } from './modules/sync/redis-adapter';
 
 const { NODE_ENV, AFFINE_ENV } = process.env;
-
-if (NODE_ENV === 'production') {
-  const traceExporter = new TraceExporter();
-  const tracing = new NodeSDK({
-    traceExporter,
-    metricReader: new PeriodicExportingMetricReader({
-      exporter: new MetricExporter(),
-    }),
-    spanProcessor: new BatchSpanProcessor(traceExporter),
-    textMapPropagator: new CompositePropagator({
-      propagators: [
-        new W3CBaggagePropagator(),
-        new W3CTraceContextPropagator(),
-      ],
-    }),
-    instrumentations: [
-      new NestInstrumentation(),
-      new ioredis.IORedisInstrumentation(),
-      new socketIO.SocketIoInstrumentation({ traceReserved: true }),
-      new gql.GraphQLInstrumentation({ mergeItems: true }),
-      new HttpInstrumentation(),
-      new PrismaInstrumentation(),
-    ],
-    serviceName: 'affine-cloud',
-  });
-
-  tracing.start();
-}
-
 const app = await NestFactory.create<NestExpressApplication>(AppModule, {
   cors: true,
   rawBody: true,

packages/backend/server/src/metrics/controller.ts

@@ -1,18 +0,0 @@
-import { Controller, Get, Res } from '@nestjs/common';
-import type { Response } from 'express';
-import { register } from 'prom-client';
-
-import { PrismaService } from '../prisma';
-
-@Controller()
-export class MetricsController {
-  constructor(private readonly prisma: PrismaService) {}
-
-  @Get('/metrics')
-  async index(@Res() res: Response): Promise<void> {
-    res.header('Content-Type', register.contentType);
-    const prismaMetrics = await this.prisma.$metrics.prometheus();
-    const appMetrics = await register.metrics();
-    res.send(appMetrics + prismaMetrics);
-  }
-}

packages/backend/server/src/metrics/index.ts

@@ -1,13 +1,3 @@
-import { Global, Module } from '@nestjs/common';
+export * from './metrics';
-
+export { start } from './opentelemetry';
-import { MetricsController } from '../metrics/controller';
+export * from './utils';
-import { Metrics } from './metrics';
-
-@Global()
-@Module({
-  providers: [Metrics],
-  exports: [Metrics],
-  controllers: [MetricsController],
-})
-export class MetricsModule {}
-export { Metrics };

packages/backend/server/src/metrics/metrics.ts

@@ -1,31 +1,76 @@
-import { Injectable, OnModuleDestroy } from '@nestjs/common';
+import opentelemetry, { Attributes, Observable } from '@opentelemetry/api';
-import { register } from 'prom-client';
 
-import { metricsCreator } from './utils';
+interface AsyncMetric {
+  ob: Observable;
+  get value(): any;
+  get attrs(): Attributes | undefined;
+}
 
-@Injectable()
+let _metrics: ReturnType<typeof createBusinessMetrics> | undefined = undefined;
-export class Metrics implements OnModuleDestroy {
+
-  onModuleDestroy(): void {
+export function getMeter(name = 'business') {
-    register.clear();
+  return opentelemetry.metrics.getMeter(name);
+}
+
+function createBusinessMetrics() {
+  const meter = getMeter();
+  const asyncMetrics: AsyncMetric[] = [];
+
+  function createGauge(name: string) {
+    let value: any;
+    let attrs: Attributes | undefined;
+    const ob = meter.createObservableGauge(name);
+    asyncMetrics.push({
+      ob,
+      get value() {
+        return value;
+      },
+      get attrs() {
+        return attrs;
+      },
+    });
+
+    return (newValue: any, newAttrs?: Attributes) => {
+      value = newValue;
+      attrs = newAttrs;
+    };
   }
 
-  socketIOEventCounter = metricsCreator.counter('socket_io_counter', ['event']);
+  const metrics = {
-  socketIOEventTimer = metricsCreator.timer('socket_io_timer', ['event']);
+    socketIOConnectionGauge: createGauge('socket_io_connection'),
-  socketIOConnectionGauge = metricsCreator.gauge(
+
-    'socket_io_connection_counter'
+    gqlRequest: meter.createCounter('gql_request'),
+    gqlError: meter.createCounter('gql_error'),
+    gqlTimer: meter.createHistogram('gql_timer'),
+
+    jwstCodecMerge: meter.createCounter('jwst_codec_merge'),
+    jwstCodecDidnotMatch: meter.createCounter('jwst_codec_didnot_match'),
+    jwstCodecFail: meter.createCounter('jwst_codec_fail'),
+
+    authCounter: meter.createCounter('auth'),
+    authFailCounter: meter.createCounter('auth_fail'),
+
+    docHistoryCounter: meter.createCounter('doc_history_created'),
+    docRecoverCounter: meter.createCounter('doc_history_recovered'),
+  };
+
+  meter.addBatchObservableCallback(
+    result => {
+      asyncMetrics.forEach(metric => {
+        result.observe(metric.ob, metric.value, metric.attrs);
+      });
+    },
+    asyncMetrics.map(({ ob }) => ob)
   );
 
-  gqlRequest = metricsCreator.counter('gql_request', ['operation']);
+  return metrics;
-  gqlError = metricsCreator.counter('gql_error', ['operation']);
-  gqlTimer = metricsCreator.timer('gql_timer', ['operation']);
-
-  jwstCodecMerge = metricsCreator.counter('jwst_codec_merge');
-  jwstCodecDidnotMatch = metricsCreator.counter('jwst_codec_didnot_match');
-  jwstCodecFail = metricsCreator.counter('jwst_codec_fail');
-
-  authCounter = metricsCreator.counter('auth');
-  authFailCounter = metricsCreator.counter('auth_fail', ['reason']);
-
-  docHistoryCounter = metricsCreator.counter('doc_history_created');
-  docRecoverCounter = metricsCreator.counter('doc_history_recovered');
 }
+
+export function registerBusinessMetrics() {
+  if (!_metrics) {
+    _metrics = createBusinessMetrics();
+  }
+
+  return _metrics;
+}
+export const metrics = registerBusinessMetrics;

packages/backend/server/src/metrics/opentelemetry.ts

@@ -0,0 +1,127 @@
+import { MetricExporter } from '@google-cloud/opentelemetry-cloud-monitoring-exporter';
+import { TraceExporter } from '@google-cloud/opentelemetry-cloud-trace-exporter';
+import {
+  CompositePropagator,
+  W3CBaggagePropagator,
+  W3CTraceContextPropagator,
+} from '@opentelemetry/core';
+import { PrometheusExporter } from '@opentelemetry/exporter-prometheus';
+import { ZipkinExporter } from '@opentelemetry/exporter-zipkin';
+import { HostMetrics } from '@opentelemetry/host-metrics';
+import { Instrumentation } from '@opentelemetry/instrumentation';
+import { GraphQLInstrumentation } from '@opentelemetry/instrumentation-graphql';
+import { HttpInstrumentation } from '@opentelemetry/instrumentation-http';
+import { IORedisInstrumentation } from '@opentelemetry/instrumentation-ioredis';
+import { NestInstrumentation } from '@opentelemetry/instrumentation-nestjs-core';
+import { SocketIoInstrumentation } from '@opentelemetry/instrumentation-socket.io';
+import {
+  ConsoleMetricExporter,
+  MetricReader,
+  PeriodicExportingMetricReader,
+} from '@opentelemetry/sdk-metrics';
+import { NodeSDK } from '@opentelemetry/sdk-node';
+import {
+  BatchSpanProcessor,
+  ConsoleSpanExporter,
+  SpanExporter,
+} from '@opentelemetry/sdk-trace-node';
+import { PrismaInstrumentation } from '@prisma/instrumentation';
+
+import { registerBusinessMetrics } from './metrics';
+
+abstract class OpentelemetryFactor {
+  abstract getMetricReader(): MetricReader;
+  abstract getSpanExporter(): SpanExporter;
+
+  getInstractions(): Instrumentation[] {
+    return [
+      new NestInstrumentation(),
+      new IORedisInstrumentation(),
+      new SocketIoInstrumentation({ traceReserved: true }),
+      new GraphQLInstrumentation({ mergeItems: true }),
+      new HttpInstrumentation(),
+      new PrismaInstrumentation(),
+    ];
+  }
+
+  create() {
+    const traceExporter = this.getSpanExporter();
+    return new NodeSDK({
+      traceExporter,
+      metricReader: this.getMetricReader(),
+      spanProcessor: new BatchSpanProcessor(traceExporter),
+      textMapPropagator: new CompositePropagator({
+        propagators: [
+          new W3CBaggagePropagator(),
+          new W3CTraceContextPropagator(),
+        ],
+      }),
+      instrumentations: this.getInstractions(),
+      serviceName: 'affine-cloud',
+    });
+  }
+}
+
+class GCloudOpentelemetryFactor extends OpentelemetryFactor {
+  override getMetricReader(): MetricReader {
+    return new PeriodicExportingMetricReader({
+      exportIntervalMillis: 30000,
+      exportTimeoutMillis: 60000,
+      exporter: new MetricExporter(),
+    });
+  }
+
+  override getSpanExporter(): SpanExporter {
+    return new TraceExporter();
+  }
+}
+
+class LocalOpentelemetryFactor extends OpentelemetryFactor {
+  override getMetricReader(): MetricReader {
+    return new PrometheusExporter();
+  }
+
+  override getSpanExporter(): SpanExporter {
+    return new ZipkinExporter();
+  }
+}
+
+class DebugOpentelemetryFactor extends OpentelemetryFactor {
+  override getMetricReader(): MetricReader {
+    return new PeriodicExportingMetricReader({
+      exporter: new ConsoleMetricExporter(),
+    });
+  }
+
+  override getSpanExporter(): SpanExporter {
+    return new ConsoleSpanExporter();
+  }
+}
+
+function createSDK() {
+  let factor: OpentelemetryFactor | null = null;
+  if (process.env.NODE_ENV === 'production') {
+    factor = new GCloudOpentelemetryFactor();
+  } else if (process.env.DEBUG_METRICS) {
+    factor = new DebugOpentelemetryFactor();
+  } else {
+    factor = new LocalOpentelemetryFactor();
+  }
+
+  return factor?.create();
+}
+
+function registerCustomMetrics() {
+  const host = new HostMetrics({ name: 'instance-host-metrics' });
+  host.start();
+}
+
+export function start() {
+  const sdk = createSDK();
+
+  if (sdk) {
+    sdk.start();
+    registerCustomMetrics();
+    registerBusinessMetrics();
+  }
+}

packages/backend/server/src/metrics/utils.ts

@@ -1,99 +1,11 @@
-import { Counter, Gauge, register, Summary } from 'prom-client';
+import { Attributes } from '@opentelemetry/api';
 
-function getOr<T>(name: string, or: () => T): T {
+import { getMeter } from './metrics';
-  return (register.getSingleMetric(name) as T) || or();
-}
-
-type LabelValues<T extends string> = Partial<Record<T, string | number>>;
-type MetricsCreator<T extends string> = (
-  value: number,
-  labels: LabelValues<T>
-) => void;
-type TimerMetricsCreator<T extends string> = (
-  labels: LabelValues<T>
-) => () => number;
-
-export const metricsCreatorGenerator = () => {
-  const counterCreator = <T extends string>(
-    name: string,
-    labelNames?: T[]
-  ): MetricsCreator<T> => {
-    const counter = getOr(
-      name,
-      () =>
-        new Counter({
-          name,
-          help: name,
-          ...(labelNames ? { labelNames } : {}),
-        })
-    );
-
-    return (value: number, labels: LabelValues<T>) => {
-      counter.inc(labels, value);
-    };
-  };
-
-  const gaugeCreator = <T extends string>(
-    name: string,
-    labelNames?: T[]
-  ): MetricsCreator<T> => {
-    const gauge = getOr(
-      name,
-      () =>
-        new Gauge({
-          name,
-          help: name,
-          ...(labelNames ? { labelNames } : {}),
-        })
-    );
-
-    return (value: number, labels: LabelValues<T>) => {
-      gauge.set(labels, value);
-    };
-  };
-
-  const timerCreator = <T extends string>(
-    name: string,
-    labelNames?: T[]
-  ): TimerMetricsCreator<T> => {
-    const summary = getOr(
-      name,
-      () =>
-        new Summary({
-          name,
-          help: name,
-          ...(labelNames ? { labelNames } : {}),
-        })
-    );
-
-    return (labels: LabelValues<T>) => {
-      const now = process.hrtime();
-
-      return () => {
-        const delta = process.hrtime(now);
-        const value = delta[0] + delta[1] / 1e9;
-
-        summary.observe(labels, value);
-        return value;
-      };
-    };
-  };
-
-  return {
-    counter: counterCreator,
-    gauge: gaugeCreator,
-    timer: timerCreator,
-  };
-};
-
-export const metricsCreator = metricsCreatorGenerator();
 
 export const CallTimer = (
   name: string,
-  labels: Record<string, any> = {}
+  attrs?: Attributes
 ): MethodDecorator => {
-  const timer = metricsCreator.timer(name, Object.keys(labels));
-
   // @ts-expect-error allow
   return (
     _target,
@@ -106,19 +18,27 @@ export const CallTimer = (
     }
 
     desc.value = function (...args: any[]) {
-      const endTimer = timer(labels);
+      const timer = getMeter().createHistogram(name, {
+        description: `function call time costs of ${name}`,
+      });
+      const start = Date.now();
+
+      const end = () => {
+        timer.record(Date.now() - start, attrs);
+      };
+
       let result: any;
       try {
         result = originalMethod.apply(this, args);
       } catch (e) {
-        endTimer();
+        end();
         throw e;
       }
 
       if (result instanceof Promise) {
-        return result.finally(endTimer);
+        return result.finally(end);
       } else {
-        endTimer();
+        end();
         return result;
       }
     };
@@ -129,10 +49,8 @@ export const CallTimer = (
 
 export const CallCounter = (
   name: string,
-  labels: Record<string, any> = {}
+  attrs?: Attributes
 ): MethodDecorator => {
-  const count = metricsCreator.counter(name, Object.keys(labels));
-
   // @ts-expect-error allow
   return (
     _target,
@@ -145,7 +63,11 @@ export const CallCounter = (
     }
 
     desc.value = function (...args: any[]) {
-      count(1, labels);
+      const count = getMeter().createCounter(name, {
+        description: `function call counter of ${name}`,
+      });
+
+      count.add(1, attrs);
       return originalMethod.apply(this, args);
     };
 

packages/backend/server/src/modules/auth/next-auth.controller.ts

@@ -23,7 +23,7 @@ import type { AuthAction, CookieOption, NextAuthOptions } from 'next-auth';
 import { AuthHandler } from 'next-auth/core';
 
 import { Config } from '../../config';
-import { Metrics } from '../../metrics/metrics';
+import { metrics } from '../../metrics';
 import { PrismaService } from '../../prisma/service';
 import { SessionService } from '../../session';
 import { AuthThrottlerGuard, Throttle } from '../../throttler';
@@ -46,7 +46,6 @@ export class NextAuthController {
     private readonly authService: AuthService,
     @Inject(NextAuthOptionsProvide)
     private readonly nextAuthOptions: NextAuthOptions,
-    private readonly metrics: Metrics,
     private readonly session: SessionService
   ) {
     // eslint-disable-next-line @typescript-eslint/no-non-null-assertion
@@ -90,7 +89,7 @@ export class NextAuthController {
       res.redirect(`/signin${query}`);
       return;
     }
-    this.metrics.authCounter(1, {});
+    metrics().authCounter.add(1);
     const [action, providerId] = req.url // start with request url
       .slice(BASE_URL.length) // make relative to baseUrl
       .replace(/\?.*/, '') // remove query part, use only path part
@@ -127,7 +126,7 @@ export class NextAuthController {
     const options = this.nextAuthOptions;
     if (req.method === 'POST' && action === 'session') {
       if (typeof req.body !== 'object' || typeof req.body.data !== 'object') {
-        this.metrics.authFailCounter(1, { reason: 'invalid_session_data' });
+        metrics().authFailCounter.add(1, { reason: 'invalid_session_data' });
         throw new BadRequestException(`Invalid new session data`);
       }
       const user = await this.updateSession(req, req.body.data);
@@ -210,7 +209,7 @@ export class NextAuthController {
 
     if (redirect?.endsWith('api/auth/error?error=AccessDenied')) {
       this.logger.log(`Early access redirect headers: ${req.headers}`);
-      this.metrics.authFailCounter(1, {
+      metrics().authFailCounter.add(1, {
         reason: 'no_early_access_permission',
       });
       if (

packages/backend/server/src/modules/doc/history.ts

@@ -6,7 +6,7 @@ import { Cron, CronExpression } from '@nestjs/schedule';
 import type { Snapshot } from '@prisma/client';
 
 import { Config } from '../../config';
-import { Metrics } from '../../metrics';
+import { metrics } from '../../metrics';
 import { PrismaService } from '../../prisma';
 import { SubscriptionStatus } from '../payment/service';
 import { Permission } from '../workspaces/types';
@@ -16,8 +16,7 @@ export class DocHistoryManager {
   private readonly logger = new Logger(DocHistoryManager.name);
   constructor(
     private readonly config: Config,
-    private readonly db: PrismaService,
+    private readonly db: PrismaService
-    private readonly metrics: Metrics
   ) {}
 
   @OnEvent('doc:manager:snapshot:beforeUpdate')
@@ -69,7 +68,7 @@ export class DocHistoryManager {
           // safe to ignore
           // only happens when duplicated history record created in multi processes
         });
-      this.metrics.docHistoryCounter(1, {});
+      metrics().docHistoryCounter.add(1, {});
       this.logger.log(
         `History created for ${snapshot.id} in workspace ${snapshot.workspaceId}.`
       );
@@ -183,7 +182,7 @@ export class DocHistoryManager {
     //  which is not the solution in CRDT.
     //  let user revert in client and update the data in sync system
     //    `await this.db.snapshot.update();`
-    this.metrics.docRecoverCounter(1, {});
+    metrics().docRecoverCounter.add(1, {});
 
     return history.timestamp;
   }

packages/backend/server/src/modules/doc/manager.ts

@@ -19,7 +19,7 @@ import {
 
 import { Cache } from '../../cache';
 import { Config } from '../../config';
-import { Metrics } from '../../metrics/metrics';
+import { metrics } from '../../metrics/metrics';
 import { PrismaService } from '../../prisma';
 import { mergeUpdatesInApplyWay as jwstMergeUpdates } from '../../storage';
 
@@ -70,7 +70,6 @@ export class DocManager implements OnModuleInit, OnModuleDestroy {
     private readonly automation: boolean,
     private readonly db: PrismaService,
     private readonly config: Config,
-    private readonly metrics: Metrics,
     private readonly cache: Cache,
     private readonly event: EventEmitter2
   ) {}
@@ -126,13 +125,13 @@ export class DocManager implements OnModuleInit, OnModuleDestroy {
       this.config.doc.manager.experimentalMergeWithJwstCodec &&
       updates.length < 100 /* avoid overloading */
     ) {
-      this.metrics.jwstCodecMerge(1, {});
+      metrics().jwstCodecMerge.add(1);
       const yjsResult = Buffer.from(encodeStateAsUpdate(doc));
       let log = false;
       try {
         const jwstResult = jwstMergeUpdates(updates);
         if (!compare(yjsResult, jwstResult)) {
-          this.metrics.jwstCodecDidnotMatch(1, {});
+          metrics().jwstCodecDidnotMatch.add(1);
           this.logger.warn(
             `jwst codec result doesn't match yjs codec result for: ${guid}`
           );
@@ -143,7 +142,7 @@ export class DocManager implements OnModuleInit, OnModuleDestroy {
           }
         }
       } catch (e) {
-        this.metrics.jwstCodecFail(1, {});
+        metrics().jwstCodecFail.add(1);
         this.logger.warn(`jwst apply update failed for ${guid}: ${e}`);
         log = true;
       } finally {

packages/backend/server/src/modules/sync/events/events.gateway.ts

@@ -11,8 +11,8 @@ import {
 import { Server, Socket } from 'socket.io';
 import { encodeStateAsUpdate, encodeStateVector } from 'yjs';
 
-import { Metrics } from '../../../metrics/metrics';
+import { metrics } from '../../../metrics';
-import { CallCounter, CallTimer } from '../../../metrics/utils';
+import { CallTimer } from '../../../metrics/utils';
 import { DocID } from '../../../utils/doc';
 import { Auth, CurrentUser } from '../../auth';
 import { DocManager } from '../../doc';
@@ -68,8 +68,7 @@ export const GatewayErrorWrapper = (): MethodDecorator => {
 const SubscribeMessage = (event: string) =>
   applyDecorators(
     GatewayErrorWrapper(),
-    CallCounter('socket_io_counter', { event }),
+    CallTimer('socket_io_event_duration', { event }),
-    CallTimer('socket_io_timer', { event }),
     RawSubscribeMessage(event)
   );
 
@@ -97,7 +96,6 @@ export class EventsGateway implements OnGatewayConnection, OnGatewayDisconnect {
 
   constructor(
     private readonly docManager: DocManager,
-    private readonly metric: Metrics,
     private readonly permissions: PermissionService
   ) {}
 
@@ -106,12 +104,12 @@ export class EventsGateway implements OnGatewayConnection, OnGatewayDisconnect {
 
   handleConnection() {
     this.connectionCount++;
-    this.metric.socketIOConnectionGauge(this.connectionCount, {});
+    metrics().socketIOConnectionGauge(this.connectionCount);
   }
 
   handleDisconnect() {
     this.connectionCount--;
-    this.metric.socketIOConnectionGauge(this.connectionCount, {});
+    metrics().socketIOConnectionGauge(this.connectionCount);
   }
 
   @Auth()

packages/backend/server/src/modules/workspaces/controller.ts

@@ -4,14 +4,13 @@ import {
   ForbiddenException,
   Get,
   Inject,
-  Logger,
   NotFoundException,
   Param,
   Res,
 } from '@nestjs/common';
 import type { Response } from 'express';
-import format from 'pretty-time';
 
+import { CallTimer } from '../../metrics';
 import { PrismaService } from '../../prisma';
 import { StorageProvide } from '../../storage';
 import { DocID } from '../../utils/doc';
@@ -23,8 +22,6 @@ import { Permission } from './types';
 
 @Controller('/api/workspaces')
 export class WorkspacesController {
-  private readonly logger = new Logger('WorkspacesController');
-
   constructor(
     @Inject(StorageProvide) private readonly storage: Storage,
     private readonly permission: PermissionService,
@@ -37,6 +34,7 @@ export class WorkspacesController {
   //
   // NOTE: because graphql can't represent a File, so we have to use REST API to get blob
   @Get('/:id/blobs/:name')
+  @CallTimer('doc_controller', { method: 'get_blob' })
   async blob(
     @Param('id') workspaceId: string,
     @Param('name') name: string,
@@ -61,13 +59,13 @@ export class WorkspacesController {
   @Get('/:id/docs/:guid')
   @Auth()
   @Publicable()
+  @CallTimer('doc_controller', { method: 'get_doc' })
   async doc(
     @CurrentUser() user: UserType | undefined,
     @Param('id') ws: string,
     @Param('guid') guid: string,
     @Res() res: Response
   ) {
-    const start = process.hrtime();
     const docId = new DocID(guid, ws);
     if (
       // if a user has the permission
@@ -104,11 +102,11 @@ export class WorkspacesController {
 
     res.setHeader('content-type', 'application/octet-stream');
     res.send(update);
-    this.logger.debug(`workspaces doc api: ${format(process.hrtime(start))}`);
   }
 
   @Get('/:id/docs/:guid/histories/:timestamp')
   @Auth()
+  @CallTimer('doc_controller', { method: 'get_history' })
   async history(
     @CurrentUser() user: UserType,
     @Param('id') ws: string,

packages/backend/server/tests/auth.spec.ts

@@ -5,7 +5,6 @@ import test from 'ava';
 
 import { ConfigModule } from '../src/config';
 import { GqlModule } from '../src/graphql.module';
-import { MetricsModule } from '../src/metrics';
 import { AuthModule } from '../src/modules/auth';
 import { AuthResolver } from '../src/modules/auth/resolver';
 import { AuthService } from '../src/modules/auth/service';
@@ -40,7 +39,6 @@ test.beforeEach(async () => {
       PrismaModule,
       GqlModule,
       AuthModule,
-      MetricsModule,
       RateLimiterModule,
     ],
   }).compile();

packages/backend/server/tests/doc.spec.ts

@@ -10,7 +10,6 @@ import { Doc as YDoc, encodeStateAsUpdate } from 'yjs';
 
 import { CacheModule } from '../src/cache';
 import { Config, ConfigModule } from '../src/config';
-import { MetricsModule } from '../src/metrics';
 import { DocManager, DocModule } from '../src/modules/doc';
 import { PrismaModule, PrismaService } from '../src/prisma';
 import { flushDB } from './utils';
@@ -19,7 +18,6 @@ const createModule = () => {
   return Test.createTestingModule({
     imports: [
       PrismaModule,
-      MetricsModule,
       CacheModule,
       EventEmitterModule.forRoot(),
       ConfigModule.forRoot(),

packages/backend/server/tests/history.spec.ts

@@ -6,7 +6,6 @@ import test from 'ava';
 import * as Sinon from 'sinon';
 
 import { ConfigModule } from '../src/config';
-import { MetricsModule } from '../src/metrics';
 import { DocHistoryManager } from '../src/modules/doc';
 import { PrismaModule, PrismaService } from '../src/prisma';
 import { flushDB } from './utils';
@@ -20,12 +19,7 @@ let db: PrismaService;
 test.beforeEach(async () => {
   await flushDB();
   m = await Test.createTestingModule({
-    imports: [
+    imports: [PrismaModule, ScheduleModule.forRoot(), ConfigModule.forRoot()],
-      PrismaModule,
-      MetricsModule,
-      ScheduleModule.forRoot(),
-      ConfigModule.forRoot(),
-    ],
     providers: [DocHistoryManager],
   }).compile();
 

packages/backend/server/tests/mailer.e2e.ts

@@ -12,7 +12,6 @@ import ava, { type TestFn } from 'ava';
 
 import { ConfigModule } from '../src/config';
 import { GqlModule } from '../src/graphql.module';
-import { MetricsModule } from '../src/metrics';
 import { AuthModule } from '../src/modules/auth';
 import { AuthService } from '../src/modules/auth/service';
 import { PrismaModule } from '../src/prisma';
@@ -44,7 +43,6 @@ test.beforeEach(async t => {
       PrismaModule,
       GqlModule,
       AuthModule,
-      MetricsModule,
       RateLimiterModule,
     ],
   }).compile();

packages/backend/server/tests/prometheus-metrics.spec.ts

@@ -1,78 +0,0 @@
-import { Test, TestingModule } from '@nestjs/testing';
-import test from 'ava';
-import { register } from 'prom-client';
-
-import { MetricsModule } from '../src/metrics';
-import { Metrics } from '../src/metrics/metrics';
-import { PrismaModule } from '../src/prisma';
-
-let metrics: Metrics;
-let module: TestingModule;
-
-test.beforeEach(async () => {
-  module = await Test.createTestingModule({
-    imports: [MetricsModule, PrismaModule],
-  }).compile();
-
-  metrics = module.get(Metrics);
-});
-
-test.afterEach.always(async () => {
-  await module.close();
-});
-
-test('should be able to increment counter', async t => {
-  metrics.socketIOEventCounter(1, { event: 'client-handshake' });
-  const socketIOCounterMetric = register.getSingleMetric('socket_io_counter');
-  t.truthy(socketIOCounterMetric);
-
-  t.truthy(
-    JSON.stringify((await socketIOCounterMetric!.get()).values) ===
-      '[{"value":1,"labels":{"event":"client-handshake"}}]'
-  );
-  t.pass();
-});
-
-test('should be able to timer', async t => {
-  let minimum: number;
-  {
-    const endTimer = metrics.socketIOEventTimer({ event: 'client-handshake' });
-    const a = performance.now();
-    await new Promise(resolve => setTimeout(resolve, 50));
-    const b = performance.now();
-    minimum = b - a;
-    endTimer();
-  }
-
-  let maximum: number;
-  {
-    const a = performance.now();
-    const endTimer = metrics.socketIOEventTimer({ event: 'client-handshake' });
-    await new Promise(resolve => setTimeout(resolve, 100));
-    endTimer();
-    const b = performance.now();
-    maximum = b - a;
-  }
-
-  const socketIOTimerMetric = register.getSingleMetric('socket_io_timer');
-  t.truthy(socketIOTimerMetric);
-
-  const observations = (await socketIOTimerMetric!.get()).values;
-
-  for (const observation of observations) {
-    if (
-      observation.labels.event === 'client-handshake' &&
-      'quantile' in observation.labels
-    ) {
-      t.truthy(
-        observation.value >= minimum / 1000,
-        'observation.value should be greater than minimum'
-      );
-      t.truthy(
-        observation.value <= maximum / 1000,
-        'observation.value should be less than maximum'
-      );
-    }
-  }
-  t.pass();
-});

yarn.lock

@@ -726,6 +726,9 @@ __metadata:
     "@node-rs/jsonwebtoken": "npm:^0.2.3"
     "@opentelemetry/api": "npm:^1.7.0"
     "@opentelemetry/core": "npm:^1.18.1"
+    "@opentelemetry/exporter-prometheus": "npm:^0.45.1"
+    "@opentelemetry/exporter-zipkin": "npm:^1.18.1"
+    "@opentelemetry/host-metrics": "npm:^0.33.2"
     "@opentelemetry/instrumentation": "npm:^0.45.1"
     "@opentelemetry/instrumentation-graphql": "npm:^0.36.0"
     "@opentelemetry/instrumentation-http": "npm:^0.45.1"
@@ -8739,6 +8742,19 @@ __metadata:
   languageName: node
   linkType: hard
 
+"@opentelemetry/exporter-prometheus@npm:^0.45.1":
+  version: 0.45.1
+  resolution: "@opentelemetry/exporter-prometheus@npm:0.45.1"
+  dependencies:
+    "@opentelemetry/core": "npm:1.18.1"
+    "@opentelemetry/resources": "npm:1.18.1"
+    "@opentelemetry/sdk-metrics": "npm:1.18.1"
+  peerDependencies:
+    "@opentelemetry/api": ^1.3.0
+  checksum: 0d9999eca90b407de1e496e00bc9d991c5bb2ccf069c53a0de9054457170426efdce833e958c53273b22957dcbfa064a5559f8c9358114dfa48c873799d4fc27
+  languageName: node
+  linkType: hard
+
 "@opentelemetry/exporter-trace-otlp-grpc@npm:0.45.1":
   version: 0.45.1
   resolution: "@opentelemetry/exporter-trace-otlp-grpc@npm:0.45.1"
@@ -8786,7 +8802,7 @@ __metadata:
   languageName: node
   linkType: hard
 
-"@opentelemetry/exporter-zipkin@npm:1.18.1":
+"@opentelemetry/exporter-zipkin@npm:1.18.1, @opentelemetry/exporter-zipkin@npm:^1.18.1":
   version: 1.18.1
   resolution: "@opentelemetry/exporter-zipkin@npm:1.18.1"
   dependencies:
@@ -8800,6 +8816,18 @@ __metadata:
   languageName: node
   linkType: hard
 
+"@opentelemetry/host-metrics@npm:^0.33.2":
+  version: 0.33.2
+  resolution: "@opentelemetry/host-metrics@npm:0.33.2"
+  dependencies:
+    "@opentelemetry/sdk-metrics": "npm:^1.8.0"
+    systeminformation: "npm:^5.0.0"
+  peerDependencies:
+    "@opentelemetry/api": ^1.3.0
+  checksum: 35140ecb2c4e97ab49873411a9e81a23238cf392112212f936f238091a6572c0802d6364ccea428cb7f0595cb28f4cbafbe6fa5c23510ebb1243cf7f398871e5
+  languageName: node
+  linkType: hard
+
 "@opentelemetry/instrumentation-graphql@npm:^0.36.0":
   version: 0.36.0
   resolution: "@opentelemetry/instrumentation-graphql@npm:0.36.0"
@@ -9013,7 +9041,7 @@ __metadata:
   languageName: node
   linkType: hard
 
-"@opentelemetry/sdk-metrics@npm:1.18.1, @opentelemetry/sdk-metrics@npm:^1.18.1":
+"@opentelemetry/sdk-metrics@npm:1.18.1, @opentelemetry/sdk-metrics@npm:^1.18.1, @opentelemetry/sdk-metrics@npm:^1.8.0":
   version: 1.18.1
   resolution: "@opentelemetry/sdk-metrics@npm:1.18.1"
   dependencies:
@@ -33454,6 +33482,16 @@ __metadata:
   languageName: node
   linkType: hard
 
+"systeminformation@npm:^5.0.0":
+  version: 5.21.17
+  resolution: "systeminformation@npm:5.21.17"
+  bin:
+    systeminformation: lib/cli.js
+  checksum: bc910730829f1b232a1ca824bb35dfb211489430a1a612fe9e0d27fd3cb73c0793a00cbff7512c782cac2f2dfd2416a7e698b410f6dd5b4f66cbf5ae458c9cab
+  conditions: (os=darwin | os=linux | os=win32 | os=freebsd | os=openbsd | os=netbsd | os=sunos | os=android)
+  languageName: node
+  linkType: hard
+
 "tabbable@npm:^6.0.1":
   version: 6.2.0
   resolution: "tabbable@npm:6.2.0"