dd-trace 4.20.0 → 4.21.0

package/packages/dd-trace/src/plugins/util/test.js

@@ -397,8 +397,9 @@ function addIntelligentTestRunnerSpanTags (
     testModuleSpan.setTag(TEST_ITR_FORCED_RUN, 'true')
   }
 
-  // If suites have been skipped we don't want to report the total coverage, as it will be wrong
-  if (testCodeCoverageLinesTotal !== undefined && !isSuitesSkipped) {
+  // This will not be reported unless the user has manually added code coverage.
+  // This is always the case for Mocha and Cucumber, but not for Jest.
+  if (testCodeCoverageLinesTotal !== undefined) {
     testSessionSpan.setTag(TEST_CODE_COVERAGE_LINES_PCT, testCodeCoverageLinesTotal)
     testModuleSpan.setTag(TEST_CODE_COVERAGE_LINES_PCT, testCodeCoverageLinesTotal)
   }

package/packages/dd-trace/src/profiler.js

@@ -8,7 +8,7 @@ process.once('beforeExit', () => { profiler.stop() })
 
 module.exports = {
   start: config => {
-    const { service, version, env, url, hostname, port, tags } = config
+    const { service, version, env, url, hostname, port, tags, repositoryUrl, commitSHA } = config
     const { enabled, sourceMap, exporters } = config.profiling
     const logger = {
       debug: (message) => log.debug(message),
@@ -17,7 +17,7 @@ module.exports = {
       error: (message) => log.error(message)
     }
 
-    profiler.start({
+    return profiler.start({
       enabled,
       service,
       version,
@@ -28,7 +28,9 @@ module.exports = {
       url,
       hostname,
       port,
-      tags
+      tags,
+      repositoryUrl,
+      commitSHA
     })
   },
 

package/packages/dd-trace/src/profiling/config.js

@@ -11,6 +11,7 @@ const WallProfiler = require('./profilers/wall')
 const SpaceProfiler = require('./profilers/space')
 const EventsProfiler = require('./profilers/events')
 const { oomExportStrategies, snapshotKinds } = require('./constants')
+const { GIT_REPOSITORY_URL, GIT_COMMIT_SHA } = require('../plugins/util/tags')
 const { tagger } = require('./tagger')
 const { isFalse, isTrue } = require('../util')
 
@@ -72,6 +73,13 @@ class Config {
       tagger.parse(options.tags),
       tagger.parse({ env, host, service, version, functionname })
     )
+
+    // Add source code integration tags if available
+    if (options.repositoryUrl && options.commitSHA) {
+      this.tags[GIT_REPOSITORY_URL] = options.repositoryUrl
+      this.tags[GIT_COMMIT_SHA] = options.commitSHA
+    }
+
     this.logger = ensureLogger(options.logger)
     const logger = this.logger
     function logExperimentalVarDeprecation (shortVarName) {

package/packages/dd-trace/src/profiling/profiler.js

@@ -23,15 +23,19 @@ class Profiler extends EventEmitter {
   }
 
   start (options) {
-    this._start(options).catch((err) => { if (options.logger) options.logger.error(err) })
-    return this
+    return this._start(options).catch((err) => {
+      if (options.logger) {
+        options.logger.error(err)
+      }
+      return false
+    })
   }
 
   async _start (options) {
-    if (this._enabled) return
+    if (this._enabled) return true
 
     const config = this._config = new Config(options)
-    if (!config.enabled) return
+    if (!config.enabled) return false
 
     this._logger = config.logger
     this._enabled = true
@@ -67,9 +71,11 @@ class Profiler extends EventEmitter {
       }
 
       this._capture(this._timeoutInterval)
+      return true
     } catch (e) {
       this._logger.error(e)
       this._stop()
+      return false
     }
   }
 

package/packages/dd-trace/src/profiling/profilers/events.js

@@ -1,5 +1,5 @@
 const { performance, constants, PerformanceObserver } = require('node:perf_hooks')
-const { END_TIMESTAMP, THREAD_NAME, threadNamePrefix } = require('./shared')
+const { END_TIMESTAMP } = require('./shared')
 const semver = require('semver')
 const { Function, Label, Line, Location, Profile, Sample, StringTable, ValueType } = require('pprof-format')
 const pprof = require('@datadog/pprof/')
@@ -14,7 +14,137 @@ const MS_TO_NS = 1000000
 // perf_hooks events, the emitted pprof file uses the type "timeline".
 const pprofValueType = 'timeline'
 const pprofValueUnit = 'nanoseconds'
-const threadName = `${threadNamePrefix} GC`
+
+function labelFromStr (stringTable, key, valStr) {
+  return new Label({ key, str: stringTable.dedup(valStr) })
+}
+
+function labelFromStrStr (stringTable, keyStr, valStr) {
+  return labelFromStr(stringTable, stringTable.dedup(keyStr), valStr)
+}
+
+class GCDecorator {
+  constructor (stringTable) {
+    this.stringTable = stringTable
+    this.reasonLabelKey = stringTable.dedup('gc reason')
+    this.kindLabels = []
+    this.reasonLabels = []
+    this.flagObj = {}
+
+    const kindLabelKey = stringTable.dedup('gc type')
+
+    // Create labels for all GC performance flags and kinds of GC
+    for (const [key, value] of Object.entries(constants)) {
+      if (key.startsWith('NODE_PERFORMANCE_GC_FLAGS_')) {
+        this.flagObj[key.substring(26).toLowerCase()] = value
+      } else if (key.startsWith('NODE_PERFORMANCE_GC_')) {
+        // It's a constant for a kind of GC
+        const kind = key.substring(20).toLowerCase()
+        this.kindLabels[value] = labelFromStr(stringTable, kindLabelKey, kind)
+      }
+    }
+  }
+
+  decorateSample (sampleInput, item) {
+    const { kind, flags } = node16 ? item.detail : item
+    sampleInput.label.push(this.kindLabels[kind])
+    const reasonLabel = this.getReasonLabel(flags)
+    if (reasonLabel) {
+      sampleInput.label.push(reasonLabel)
+    }
+  }
+
+  getReasonLabel (flags) {
+    if (flags === 0) {
+      return null
+    }
+    let reasonLabel = this.reasonLabels[flags]
+    if (!reasonLabel) {
+      const reasons = []
+      for (const [key, value] of Object.entries(this.flagObj)) {
+        if (value & flags) {
+          reasons.push(key)
+        }
+      }
+      const reasonStr = reasons.join(',')
+      reasonLabel = labelFromStr(this.stringTable, this.reasonLabelKey, reasonStr)
+      this.reasonLabels[flags] = reasonLabel
+    }
+    return reasonLabel
+  }
+}
+
+class DNSDecorator {
+  constructor (stringTable) {
+    this.stringTable = stringTable
+    this.operationNameLabelKey = stringTable.dedup('operation')
+    this.hostLabelKey = stringTable.dedup('host')
+    this.addressLabelKey = stringTable.dedup('address')
+    this.portLabelKey = stringTable.dedup('port')
+  }
+
+  decorateSample (sampleInput, item) {
+    const labels = sampleInput.label
+    const stringTable = this.stringTable
+    function addLabel (labelNameKey, labelValue) {
+      labels.push(labelFromStr(stringTable, labelNameKey, labelValue))
+    }
+    const op = item.name
+    addLabel(this.operationNameLabelKey, item.name)
+    const detail = item.detail
+    switch (op) {
+      case 'lookup':
+        addLabel(this.hostLabelKey, detail.hostname)
+        break
+      case 'lookupService':
+        addLabel(this.addressLabelKey, detail.host)
+        labels.push(new Label({ key: this.portLabelKey, num: detail.port }))
+        break
+      case 'getHostByAddr':
+        addLabel(this.addressLabelKey, detail.host)
+        break
+      default:
+        if (op.startsWith('query')) {
+          addLabel(this.hostLabelKey, detail.host)
+        }
+    }
+  }
+}
+
+class NetDecorator {
+  constructor (stringTable) {
+    this.stringTable = stringTable
+    this.operationNameLabelKey = stringTable.dedup('operation')
+    this.hostLabelKey = stringTable.dedup('host')
+    this.portLabelKey = stringTable.dedup('port')
+  }
+
+  decorateSample (sampleInput, item) {
+    const labels = sampleInput.label
+    const stringTable = this.stringTable
+    function addLabel (labelNameKey, labelValue) {
+      labels.push(labelFromStr(stringTable, labelNameKey, labelValue))
+    }
+    const op = item.name
+    addLabel(this.operationNameLabelKey, op)
+    if (op === 'connect') {
+      const detail = item.detail
+      addLabel(this.hostLabelKey, detail.host)
+      labels.push(new Label({ key: this.portLabelKey, num: detail.port }))
+    }
+  }
+}
+
+// Keys correspond to PerformanceEntry.entryType, values are constructor
+// functions for type-specific decorators.
+const decoratorTypes = {
+  gc: GCDecorator
+}
+// Needs at least node 16 for DNS and Net
+if (node16) {
+  decoratorTypes.dns = DNSDecorator
+  decoratorTypes.net = NetDecorator
+}
 
 /**
  * This class generates pprof files with timeline events sourced from Node.js
@@ -35,8 +165,7 @@ class EventsProfiler {
     if (!this._observer) {
       this._observer = new PerformanceObserver(add.bind(this))
     }
-    // Currently only support GC
-    this._observer.observe({ entryTypes: ['gc'] })
+    this._observer.observe({ entryTypes: Object.keys(decoratorTypes) })
   }
 
   stop () {
@@ -52,91 +181,55 @@ class EventsProfiler {
     }
 
     const stringTable = new StringTable()
-    const timestampLabelKey = stringTable.dedup(END_TIMESTAMP)
-    const kindLabelKey = stringTable.dedup('gc type')
-    const reasonLabelKey = stringTable.dedup('gc reason')
-    const kindLabels = []
-    const reasonLabels = []
     const locations = []
     const functions = []
-    const locationsPerKind = []
-    const flagObj = {}
 
-    function labelFromStr (key, valStr) {
-      return new Label({ key, str: stringTable.dedup(valStr) })
-    }
+    // A synthetic single-frame location to serve as the location for timeline
+    // samples. We need these as the profiling backend (mimicking official pprof
+    // tool's behavior) ignores these.
+    const locationId = (() => {
+      const fn = new Function({ id: functions.length + 1, name: stringTable.dedup('') })
+      functions.push(fn)
+      const line = new Line({ functionId: fn.id })
+      const location = new Location({ id: locations.length + 1, line: [line] })
+      locations.push(location)
+      return [location.id]
+    })()
 
-    function labelFromStrStr (keyStr, valStr) {
-      return labelFromStr(stringTable.dedup(keyStr), valStr)
-    }
-
-    // Create labels for all GC performance flags and kinds of GC
-    for (const [key, value] of Object.entries(constants)) {
-      if (key.startsWith('NODE_PERFORMANCE_GC_FLAGS_')) {
-        flagObj[key.substring(26).toLowerCase()] = value
-      } else if (key.startsWith('NODE_PERFORMANCE_GC_')) {
-        // It's a constant for a kind of GC
-        const kind = key.substring(20).toLowerCase()
-        kindLabels[value] = labelFromStr(kindLabelKey, kind)
-        // Construct a single-frame "location" too
-        const fn = new Function({ id: functions.length + 1, name: stringTable.dedup(`${kind} GC`) })
-        functions.push(fn)
-        const line = new Line({ functionId: fn.id })
-        const location = new Location({ id: locations.length + 1, line: [line] })
-        locations.push(location)
-        locationsPerKind[value] = [location.id]
-      }
-    }
-
-    const gcEventLabel = labelFromStrStr('event', 'gc')
-    const threadLabel = labelFromStrStr(THREAD_NAME, threadName)
-
-    function getReasonLabel (flags) {
-      if (flags === 0) {
-        return null
-      }
-      let reasonLabel = reasonLabels[flags]
-      if (!reasonLabel) {
-        const reasons = []
-        for (const [key, value] of Object.entries(flagObj)) {
-          if (value & flags) {
-            reasons.push(key)
-          }
-        }
-        const reasonStr = reasons.join(',')
-        reasonLabel = labelFromStr(reasonLabelKey, reasonStr)
-        reasonLabels[flags] = reasonLabel
-      }
-      return reasonLabel
+    const decorators = {}
+    for (const [eventType, DecoratorCtor] of Object.entries(decoratorTypes)) {
+      const decorator = new DecoratorCtor(stringTable)
+      decorator.eventTypeLabel = labelFromStrStr(stringTable, 'event', eventType)
+      decorators[eventType] = decorator
     }
+    const timestampLabelKey = stringTable.dedup(END_TIMESTAMP)
 
     let durationFrom = Number.POSITIVE_INFINITY
     let durationTo = 0
     const dateOffset = BigInt(Math.round(performance.timeOrigin * MS_TO_NS))
 
     const samples = this.entries.map((item) => {
+      const decorator = decorators[item.entryType]
+      if (!decorator) {
+        // Shouldn't happen but it's better to not rely on observer only getting
+        // requested event types.
+        return null
+      }
       const { startTime, duration } = item
-      const { kind, flags } = node16 ? item.detail : item
       const endTime = startTime + duration
       if (durationFrom > startTime) durationFrom = startTime
       if (durationTo < endTime) durationTo = endTime
-      const labels = [
-        gcEventLabel,
-        threadLabel,
-        new Label({ key: timestampLabelKey, num: dateOffset + BigInt(Math.round(endTime * MS_TO_NS)) }),
-        kindLabels[kind]
-      ]
-      const reasonLabel = getReasonLabel(flags)
-      if (reasonLabel) {
-        labels.push(reasonLabel)
-      }
-      const sample = new Sample({
+      const sampleInput = {
         value: [Math.round(duration * MS_TO_NS)],
-        label: labels,
-        locationId: locationsPerKind[kind]
-      })
-      return sample
-    })
+        locationId,
+        label: [
+          decorator.eventTypeLabel,
+          new Label({ key: timestampLabelKey, num: dateOffset + BigInt(Math.round(endTime * MS_TO_NS)) })
+        ]
+      }
+      decorator.decorateSample(sampleInput, item)
+      return new Sample(sampleInput)
+    }).filter(v => v)
 
     this.entries = []
 

package/packages/dd-trace/src/proxy.js

@@ -10,6 +10,7 @@ const PluginManager = require('./plugin_manager')
 const remoteConfig = require('./appsec/remote_config')
 const AppsecSdk = require('./appsec/sdk')
 const dogstatsd = require('./dogstatsd')
+const spanleak = require('./spanleak')
 
 class Tracer extends NoopProxy {
   constructor () {
@@ -37,6 +38,15 @@ class Tracer extends NoopProxy {
         }, 10 * 1000).unref()
       }
 
+      if (config.spanLeakDebug > 0) {
+        if (config.spanLeakDebug === spanleak.MODES.LOG) {
+          spanleak.enableLogging()
+        } else if (config.spanLeakDebug === spanleak.MODES.GC_AND_LOG) {
+          spanleak.enableGarbageCollection()
+        }
+        spanleak.startScrubber()
+      }
+
       if (config.remoteConfig.enabled && !config.isCiVisibility) {
         const rc = remoteConfig.enable(config)
 
@@ -62,11 +72,14 @@ class Tracer extends NoopProxy {
         // do not stop tracer initialization if the profiler fails to be imported
         try {
           const profiler = require('./profiler')
-          profiler.start(config)
+          this._profilerStarted = profiler.start(config)
         } catch (e) {
           log.error(e)
         }
       }
+      if (!this._profilerStarted) {
+        this._profilerStarted = Promise.resolve(false)
+      }
 
       if (config.runtimeMetrics) {
         runtimeMetrics.start(config)
@@ -104,6 +117,13 @@ class Tracer extends NoopProxy {
     return this
   }
 
+  profilerStarted () {
+    if (!this._profilerStarted) {
+      throw new Error('profilerStarted() must be called after init()')
+    }
+    return this._profilerStarted
+  }
+
   use () {
     this._pluginManager.configurePlugin(...arguments)
     return this

package/packages/dd-trace/src/service-naming/schemas/v0/storage.js

@@ -37,6 +37,11 @@ const redisConfig = {
 
 const storage = {
   client: {
+    aerospike: {
+      opName: () => 'aerospike.command',
+      serviceName: ({ tracerService, pluginConfig }) =>
+        pluginConfig.service || `${tracerService}-aerospike`
+    },
     'cassandra-driver': {
       opName: () => 'cassandra.query',
       serviceName: ({ tracerService, pluginConfig, system }) =>

package/packages/dd-trace/src/service-naming/schemas/v1/storage.js

@@ -22,6 +22,10 @@ function withFunction ({ tracerService, pluginConfig, params }) {
 
 const storage = {
   client: {
+    aerospike: {
+      opName: () => 'aerospike.command',
+      serviceName: configWithFallback
+    },
     'cassandra-driver': {
       opName: () => 'cassandra.query',
       serviceName: configWithFallback

package/packages/dd-trace/src/spanleak.js

@@ -0,0 +1,98 @@
+'use strict'
+
+/* eslint-disable no-console */
+
+const SortedSet = require('tlhunter-sorted-set')
+
+const INTERVAL = 1000 // look for expired spans every 1s
+const LIFETIME = 60 * 1000 // all spans have a max lifetime of 1m
+
+const MODES = {
+  DISABLED: 0,
+  // METRICS_ONLY
+  LOG: 1,
+  GC_AND_LOG: 2
+  // GC
+}
+
+module.exports.MODES = MODES
+
+const spans = new SortedSet()
+
+// TODO: should these also be delivered as runtime metrics?
+
+// const registry = new FinalizationRegistry(name => {
+//   spans.del(span) // there is no span
+// })
+
+let interval
+let mode = MODES.DISABLED
+
+module.exports.disable = function () {
+  mode = MODES.DISABLED
+}
+
+module.exports.enableLogging = function () {
+  mode = MODES.LOG
+}
+
+module.exports.enableGarbageCollection = function () {
+  mode = MODES.GC_AND_LOG
+}
+
+module.exports.startScrubber = function () {
+  if (!isEnabled()) return
+
+  interval = setInterval(() => {
+    const now = Date.now()
+    const expired = spans.rangeByScore(0, now)
+
+    if (!expired.length) return
+
+    const gc = isGarbageCollecting()
+
+    const expirationsByType = Object.create(null) // { [spanType]: count }
+
+    for (const wrapped of expired) {
+      spans.del(wrapped)
+      const span = wrapped.deref()
+
+      if (!span) continue // span has already been garbage collected
+
+      // TODO: Should we also do things like record the route to help users debug leaks?
+      if (!expirationsByType[span._name]) expirationsByType[span._name] = 0
+      expirationsByType[span._name]++
+
+      if (!gc) continue // everything after this point is related to manual GC
+
+      // TODO: what else can we do to alleviate memory usage
+      span.context()._tags = Object.create(null)
+    }
+
+    console.log('expired spans:' +
+      Object.keys(expirationsByType).reduce((a, c) => `${a} ${c}: ${expirationsByType[c]}`, ''))
+  }, INTERVAL)
+}
+
+module.exports.stopScrubber = function () {
+  clearInterval(interval)
+}
+
+module.exports.addSpan = function (span) {
+  if (!isEnabled()) return
+
+  const now = Date.now()
+  const expiration = now + LIFETIME
+  // eslint-disable-next-line no-undef
+  const wrapped = new WeakRef(span)
+  spans.add(wrapped, expiration)
+  // registry.register(span, span._name)
+}
+
+function isEnabled () {
+  return mode > MODES.DISABLED
+}
+
+function isGarbageCollecting () {
+  return mode >= MODES.GC_AND_LOG
+}

package/packages/dd-trace/src/startup-log.js

@@ -6,6 +6,7 @@ const os = require('os')
 const { inspect } = require('util')
 const tracerVersion = require('../../../package.json').version
 
+const errors = {}
 let config
 let pluginManager
 let samplingRules = []
@@ -89,6 +90,10 @@ function startupLog ({ agentError } = {}) {
   info('DATADOG TRACER CONFIGURATION - ' + out)
   if (agentError) {
     warn('DATADOG TRACER DIAGNOSTIC - Agent Error: ' + agentError.message)
+    errors.agentError = {
+      code: agentError.code ? agentError.code : '',
+      message: `Agent Error:${agentError.message}`
+    }
   }
 
   config = undefined
@@ -112,5 +117,6 @@ module.exports = {
   startupLog,
   setStartupLogConfig,
   setStartupLogPluginManager,
-  setSamplingRules
+  setSamplingRules,
+  errors
 }

package/packages/dd-trace/src/telemetry/dependencies.js

@@ -6,6 +6,7 @@ const requirePackageJson = require('../require-package-json')
 const { sendData } = require('./send-data')
 const dc = require('dc-polyfill')
 const { fileURLToPath } = require('url')
+const { isTrue } = require('../../src/util')
 
 const savedDependenciesToSend = new Set()
 const detectedDependencyKeys = new Set()
@@ -14,20 +15,57 @@ const detectedDependencyVersions = new Set()
 const FILE_URI_START = `file://`
 const moduleLoadStartChannel = dc.channel('dd-trace:moduleLoadStart')
 
-let immediate, config, application, host
+let immediate, config, application, host, initialLoad
 let isFirstModule = true
+let getRetryData
+let updateRetryData
 
+function createBatchPayload (payload) {
+  const batchPayload = []
+  payload.map(item => {
+    batchPayload.push({
+      request_type: item.reqType,
+      payload: item.payload
+    })
+  })
+
+  return batchPayload
+}
 function waitAndSend (config, application, host) {
   if (!immediate) {
     immediate = setImmediate(() => {
       immediate = null
       if (savedDependenciesToSend.size > 0) {
-        const dependencies = Array.from(savedDependenciesToSend.values()).splice(0, 1000).map(pair => {
-          savedDependenciesToSend.delete(pair)
-          const [name, version] = pair.split(' ')
-          return { name, version }
-        })
-        sendData(config, application, host, 'app-dependencies-loaded', { dependencies })
+        const dependencies = Array.from(savedDependenciesToSend.values())
+          // if a depencdency is from the initial load, *always* send the event
+          // Otherwise, only send if dependencyCollection is enabled
+          .filter(dep => {
+            const initialLoadModule = isTrue(dep.split(' ')[2])
+            const sendModule = initialLoadModule || (config.telemetry?.dependencyCollection)
+
+            if (!sendModule) savedDependenciesToSend.delete(dep) // we'll never send it
+            return sendModule
+          })
+          .splice(0, 2000) // v2 documentation specifies up to 2000 dependencies can be sent at once
+          .map(pair => {
+            savedDependenciesToSend.delete(pair)
+            const [name, version] = pair.split(' ')
+            return { name, version }
+          })
+        let currPayload
+        const retryData = getRetryData()
+        if (retryData) {
+          currPayload = { reqType: 'app-dependencies-loaded', payload: { dependencies } }
+        } else {
+          if (!dependencies.length) return // no retry data and no dependencies, nothing to send
+          currPayload = { dependencies }
+        }
+
+        const payload = retryData ? createBatchPayload([currPayload, retryData]) : currPayload
+        const reqType = retryData ? 'message-batch' : 'app-dependencies-loaded'
+
+        sendData(config, application, host, reqType, payload, updateRetryData)
+
         if (savedDependenciesToSend.size > 0) {
           waitAndSend(config, application, host)
         }
@@ -76,7 +114,7 @@ function onModuleLoad (data) {
             const dependencyAndVersion = `${name} ${version}`
 
             if (!detectedDependencyVersions.has(dependencyAndVersion)) {
-              savedDependenciesToSend.add(dependencyAndVersion)
+              savedDependenciesToSend.add(`${dependencyAndVersion} ${initialLoad}`)
               detectedDependencyVersions.add(dependencyAndVersion)
 
               waitAndSend(config, application, host)
@@ -89,11 +127,19 @@ function onModuleLoad (data) {
     }
   }
 }
-function start (_config, _application, _host) {
+function start (_config = {}, _application, _host, getRetryDataFunction, updateRetryDatafunction) {
   config = _config
   application = _application
   host = _host
+  initialLoad = true
+  getRetryData = getRetryDataFunction
+  updateRetryData = updateRetryDatafunction
   moduleLoadStartChannel.subscribe(onModuleLoad)
+
+  // try and capture intially loaded modules in the first tick
+  // since, ideally, the tracer (and this module) should be loaded first,
+  // this should capture any first-tick dependencies
+  queueMicrotask(() => { initialLoad = false })
 }
 
 function isDependency (filename, request) {