👷[RUM-11488] enforce specifying an expiration date for temporary telemetry (#3825)

* 👷add eslint rule to enforce `monitor-until` comments

* 👷define expiration date or expire exiting telemetry

* 👷add eslint rule to detect expired monitor-until comments

* 👷add a scheduled job to check expired telemetry

* 👌remove unnecessary ssh eval + report

* 👌adjust replay telemetry expiration

Author: bcaudan

.gitlab-ci.yml

@@ -586,3 +586,30 @@ bump-chrome-version-scheduled-failure:
     - postmessage "#browser-sdk-deploy" "$MESSAGE_TEXT"
   dependencies:
     - bump-chrome-version-scheduled
+
+########################################################################################################################
+# Check expired telemetry
+########################################################################################################################
+
+check-expired-telemetry-scheduled:
+  stage: task
+  extends: .base-configuration
+  only:
+    variables:
+      - $TARGET_TASK_NAME == "check-expired-telemetry"
+  script:
+    - yarn
+    - yarn build
+    - MONITOR_UNTIL_COMMENT_EXPIRED_LEVEL=error yarn lint
+
+check-expired-telemetry-scheduled-failure:
+  extends: .prepare_notification
+  only:
+    variables:
+      - $TARGET_TASK_NAME == "check-expired-telemetry"
+  when: on_failure
+  script:
+    - 'MESSAGE_TEXT=":fire: [*$CI_PROJECT_NAME*] <$BUILD_URL|Expired telemetry detected> :fire:"'
+    - postmessage "#rum-browser-sdk-ops" "$MESSAGE_TEXT"
+  dependencies:
+    - check-expired-telemetry-scheduled

eslint-local-rules/index.js

@@ -18,4 +18,5 @@ module.exports = {
   'disallow-non-scripts': require('./disallowNonScripts'),
   'enforce-prod-deps-imports': require('./enforceProdDepsImports.js'),
   'secure-command-execution': require('./secureCommandExecution'),
+  ...require('./monitorUntilCommentRules'),
 }

eslint-local-rules/monitorUntilCommentRules.js

@@ -0,0 +1,62 @@
+const METHODS_TO_CHECK = ['addTelemetryDebug', 'addTelemetryMetrics']
+const MONITOR_COMMENT_FORMAT = /^\s*monitor-until: (\d{4}-\d{2}-\d{2}|forever)/
+
+module.exports = {
+  'enforce-monitor-until-comment': {
+    meta: {
+      docs: {
+        description:
+          'Force to specify an expiration date to telemetry debug and metrics events in order to clean them regularly',
+        recommended: false,
+      },
+      schema: [],
+    },
+    create(context) {
+      const sourceCode = context.getSourceCode()
+
+      return {
+        CallExpression(node) {
+          const methodName = node.callee.name
+          if (!METHODS_TO_CHECK.includes(methodName)) {
+            return
+          }
+
+          const monitorComment = sourceCode
+            .getCommentsBefore(node)
+            .find((comment) => MONITOR_COMMENT_FORMAT.test(comment.value))
+
+          if (!monitorComment) {
+            context.report(node, 'Missing `// monitor-until: YYYY-MM-DD` or `// monitor-until: forever` comment')
+          }
+        },
+      }
+    },
+  },
+  'monitor-until-comment-expired': {
+    meta: {
+      docs: {
+        description: 'Report expired monitor-until comments',
+        recommended: false,
+      },
+      schema: [],
+    },
+    create(context) {
+      return {
+        Program() {
+          const now = new Date()
+          const comments = context.getSourceCode().getAllComments()
+          comments.forEach((comment) => {
+            const monitorCommentMatch = comment.value.match(MONITOR_COMMENT_FORMAT)
+            if (!monitorCommentMatch || monitorCommentMatch[1] === 'forever') {
+              return
+            }
+
+            if (new Date(monitorCommentMatch[1]) < now) {
+              context.report(comment, `Expired: ${comment.value}`)
+            }
+          })
+        },
+      }
+    },
+  },
+}

eslint.config.mjs

@@ -11,6 +11,7 @@ import globals from 'globals'
 import eslintLocalRules from './eslint-local-rules/index.js'
 
 const SPEC_FILES = '**/*.{spec,specHelper}.{ts,tsx,js}'
+const MONITOR_UNTIL_COMMENT_EXPIRED_LEVEL = process.env.MONITOR_UNTIL_COMMENT_EXPIRED_LEVEL || 'warn'
 
 // eslint-disable-next-line import/no-default-export
 export default tseslint.config(
@@ -357,6 +358,8 @@ export default tseslint.config(
     files: ['packages/*/src/**/*.ts'],
     ignores: [SPEC_FILES],
     rules: {
+      'local-rules/enforce-monitor-until-comment': 'error',
+      'local-rules/monitor-until-comment-expired': MONITOR_UNTIL_COMMENT_EXPIRED_LEVEL,
       'local-rules/disallow-side-effects': 'error',
       'local-rules/disallow-zone-js-patched-values': 'error',
       'local-rules/disallow-url-constructor-patched-values': 'error',

packages/core/src/domain/allowedTrackingOrigins.ts

@@ -19,6 +19,7 @@ export function isAllowedTrackingOrigins(
       display.warn(WARN_DOES_NOT_HAVE_ALLOWED_TRACKING_ORIGIN)
 
       const extensionUrl = extractExtensionUrlFromStack(errorStack)
+      // monitor-until: 2026-01-01
       addTelemetryDebug(WARN_DOES_NOT_HAVE_ALLOWED_TRACKING_ORIGIN, {
         extensionUrl: extensionUrl || 'unknown',
       })

packages/core/src/domain/session/sessionManager.ts

@@ -174,6 +174,7 @@ function trackResume(configuration: Configuration, cb: () => void) {
 
 async function reportUnexpectedSessionState() {
   const rawSession = retrieveSessionCookie()
+  // monitor-until: forever, could be handy to troubleshoot issues until session manager rework
   addTelemetryDebug('Unexpected session state', {
     session: rawSession,
     isSyntheticsTest: isSyntheticsTest(),
@@ -212,6 +213,7 @@ function detectSessionIdChange(configuration: Configuration, initialSessionState
         const time = dateNow() - sdkInitTime
         getSessionCookies()
           .then((cookie) => {
+            // monitor-until: 2025-10-01, after investigation done
             addTelemetryDebug('Session cookie changed', {
               time,
               session_age: sessionAge,

packages/core/src/domain/session/sessionStoreOperations.ts

@@ -2,7 +2,6 @@ import { setTimeout } from '../../tools/timer'
 import { generateUUID } from '../../tools/utils/stringUtils'
 import type { TimeStamp } from '../../tools/utils/timeUtils'
 import { elapsed, ONE_SECOND, timeStampNow } from '../../tools/utils/timeUtils'
-import { addTelemetryDebug } from '../telemetry'
 import type { SessionStoreStrategy } from './storeStrategies/sessionStoreStrategy'
 import type { SessionState } from './sessionState'
 import { expandSessionState, isSessionInExpiredState } from './sessionState'
@@ -46,9 +45,6 @@ export function processSessionStoreOperations(
     return
   }
   if (isLockEnabled && numberOfRetries >= LOCK_MAX_TRIES) {
-    addTelemetryDebug('Aborted session operation after max lock retries', {
-      currentStore: retrieveStore(),
-    })
     next(sessionStoreStrategy)
     return
   }

packages/logs/src/boot/startLogs.ts

@@ -80,7 +80,7 @@ export function startLogs(
   const accountContext = startAccountContext(hooks, configuration, LOGS_STORAGE_KEY)
   const userContext = startUserContext(hooks, configuration, session, LOGS_STORAGE_KEY)
   const globalContext = startGlobalContext(hooks, configuration, LOGS_STORAGE_KEY, false)
-  const { stop } = startRUMInternalContext(hooks)
+  startRUMInternalContext(hooks)
 
   startNetworkErrorCollection(configuration, lifeCycle)
   startRuntimeErrorCollection(configuration, lifeCycle, bufferedDataObservable)
@@ -114,7 +114,6 @@ export function startLogs(
     userContext,
     stop: () => {
       cleanupTasks.forEach((task) => task())
-      stop()
     },
   }
 }

packages/logs/src/domain/contexts/rumInternalContext.spec.ts

@@ -1,23 +1,21 @@
 import type { RelativeTime } from '@datadog/browser-core'
 import { HookNames } from '@datadog/browser-core'
-import { mockSyntheticsWorkerValues, startMockTelemetry } from '@datadog/browser-core/test'
+import { mockSyntheticsWorkerValues } from '@datadog/browser-core/test'
 import type { Hooks } from '../hooks'
 import { createHooks } from '../hooks'
 import { startRUMInternalContext } from './rumInternalContext'
 
 describe('startRUMInternalContext', () => {
   let hooks: Hooks
-  let stopRUMInternalContext: () => void
 
   beforeEach(() => {
     hooks = createHooks()
-    ;({ stop: stopRUMInternalContext } = startRUMInternalContext(hooks))
+    startRUMInternalContext(hooks)
   })
 
   afterEach(() => {
     delete window.DD_RUM
     delete window.DD_RUM_SYNTHETICS
-    stopRUMInternalContext()
   })
 
   describe('assemble hook', () => {
@@ -61,33 +59,6 @@ describe('startRUMInternalContext', () => {
         })
         expect(defaultLogsEventAttributes).toEqual({ foo: 'bar' })
       })
-
-      it('adds a telemetry debug event when RUM has not been injected yet', async () => {
-        const telemetry = startMockTelemetry()
-        hooks.triggerHook(HookNames.Assemble, {
-          startTime: 0 as RelativeTime,
-        })
-        expect(await telemetry.getEvents()).toEqual([
-          jasmine.objectContaining({
-            message: 'Logs sent before RUM is injected by the synthetics worker',
-            status: 'debug',
-            type: 'log',
-            testId: 'test-id',
-            resultId: 'result-id',
-          }),
-        ])
-      })
-
-      it('adds the telemetry debug event only once', async () => {
-        const telemetry = startMockTelemetry()
-        hooks.triggerHook(HookNames.Assemble, {
-          startTime: 0 as RelativeTime,
-        })
-        hooks.triggerHook(HookNames.Assemble, {
-          startTime: 0 as RelativeTime,
-        })
-        expect((await telemetry.getEvents()).length).toEqual(1)
-      })
     })
   })
 

packages/logs/src/domain/contexts/rumInternalContext.ts

@@ -1,13 +1,5 @@
 import type { RelativeTime, RumInternalContext } from '@datadog/browser-core'
-import {
-  globalObject,
-  willSyntheticsInjectRum,
-  addTelemetryDebug,
-  getSyntheticsTestId,
-  getSyntheticsResultId,
-  HookNames,
-  SKIPPED,
-} from '@datadog/browser-core'
+import { globalObject, willSyntheticsInjectRum, HookNames, SKIPPED } from '@datadog/browser-core'
 import type { Hooks } from '../hooks'
 
 interface Rum {
@@ -21,7 +13,6 @@ interface BrowserWindow {
 
 export function startRUMInternalContext(hooks: Hooks) {
   const browserWindow = globalObject as BrowserWindow
-  let logsSentBeforeRumInjectionTelemetryAdded = false
 
   hooks.register(HookNames.Assemble, ({ startTime }) => {
     const internalContext = getRUMInternalContext(startTime)
@@ -54,25 +45,11 @@ export function startRUMInternalContext(hooks: Hooks) {
     if (rumContext) {
       return rumContext
     }
-
-    if (willSyntheticsInjectRumResult && !logsSentBeforeRumInjectionTelemetryAdded) {
-      logsSentBeforeRumInjectionTelemetryAdded = true
-      addTelemetryDebug('Logs sent before RUM is injected by the synthetics worker', {
-        testId: getSyntheticsTestId(),
-        resultId: getSyntheticsResultId(),
-      })
-    }
   }
 
   function getInternalContextFromRumGlobal(startTime?: RelativeTime, rumGlobal?: Rum): RumInternalContext | undefined {
     if (rumGlobal && rumGlobal.getInternalContext) {
       return rumGlobal.getInternalContext(startTime)
     }
   }
-
-  return {
-    stop: () => {
-      logsSentBeforeRumInjectionTelemetryAdded = false
-    },
-  }
 }