Flink运行时之客户端提交作业图-下

文章目录
  1. 1. submitJob方法分析

submitJob方法分析

JobClientActor通过向JobManager的Actor发送SubmitJob消息来提交Job,JobManager接收到消息对象之后,构建一个JobInfo对象以封装Job的基本信息,然后将这两个对象传递给submitJob方法:

1
2
3
4
5
case SubmitJob(jobGraph, listeningBehaviour) =>
val client = sender()
val jobInfo = new JobInfo(client, listeningBehaviour, System.currentTimeMillis(),
jobGraph.getSessionTimeout)
submitJob(jobGraph, jobInfo)

我们会以submitJob的关键方法调用来串讲其主要逻辑。首先判断jobGraph参数,如果为空则直接回应JobResultFailure消息:

1
2
3
4
5
6
7
if (jobGraph == null) {
jobInfo.client ! decorateMessage(JobResultFailure(
new SerializedThrowable(
new JobSubmissionException(null, "JobGraph must not be null.")
)
))
}

接着,向类库缓存管理器注册该Job相关的库文件、类路径:

1
2
libraryCacheManager.registerJob(jobGraph.getJobID, jobGraph.getUserJarBlobKeys,
jobGraph.getClasspaths)

必须确保该步骤率先成功执行,因为一旦后续产生任何异常才可以确保上传的类库和Jar等被成功从类库缓存管理器中移除。从这开始的整个代码段都被包裹在try语句块中,一旦捕获到任何异常,会通过libraryCacheManager的unregisterJob方法将相关Jar文件删除:

1
2
3
4
catch { case t: Throwable =>
libraryCacheManager.unregisterJob(jobId)
//...
}

接下来是获得用户代码的类加载器classLoader以及发生失败时的重启策略restartStrategy:

1
2
3
4
5
6
val userCodeLoader = libraryCacheManager.getClassLoader(jobGraph.getJobID)
val restartStrategy = Option(jobGraph.getRestartStrategyConfiguration())
.map(RestartStrategyFactory.createRestartStrategy(_)) match {
case Some(strategy) => strategy
case None => defaultRestartStrategy
}

接着,获得执行图ExecutionGraph对象的实例。首先尝试从缓存中查找,如果缓存中存在则直接返回,否则直接创建然后加入缓存:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
executionGraph = currentJobs.get(jobGraph.getJobID) match {
case Some((graph, currentJobInfo)) =>
currentJobInfo.setLastActive()
graph
case None =>
val graph = new ExecutionGraph(
executionContext,
jobGraph.getJobID,
jobGraph.getName,
jobGraph.getJobConfiguration,
timeout,
restartStrategy,
jobGraph.getUserJarBlobKeys,
jobGraph.getClasspaths,
userCodeLoader)
currentJobs.put(jobGraph.getJobID, (graph, jobInfo))
graph
}

获得了executionGraph之后会对其相关属性进行设置,这些属性包括调度模式、是否允许被加入调度队列、计划的Json格式表示。

1
2
3
executionGraph.setScheduleMode(jobGraph.getScheduleMode())
executionGraph.setQueuedSchedulingAllowed(jobGraph.getAllowQueuedScheduling())
executionGraph.setJsonPlan(JsonPlanGenerator.generatePlan(jobGraph))

接下来初始化JobVertex的一些属性:

1
2
3
4
5
6
7
8
val numSlots = scheduler.getTotalNumberOfSlots()
for (vertex <- jobGraph.getVertices.asScala) {
val executableClass = vertex.getInvokableClassName
if (vertex.getParallelism() == ExecutionConfig.PARALLELISM_AUTO_MAX) {
vertex.setParallelism(numSlots)
}
vertex.initializeOnMaster(userCodeLoader)
}

获得JobGraph中从source开始的按照拓扑顺序排序的顶点集合,然后将该集合附加到ExecutionGraph上,附加的过程完成了很多事情,我们后续进行分析:

1
2
val sortedTopology = jobGraph.getVerticesSortedTopologicallyFromSources()
executionGraph.attachJobGraph(sortedTopology)

接下来将快照配置和检查点配置的信息写入ExecutionGraph:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
val snapshotSettings = jobGraph.getSnapshotSettings
if (snapshotSettings != null) {
val jobId = jobGraph.getJobID()
val idToVertex: JobVertexID => ExecutionJobVertex = id => {
val vertex = executionGraph.getJobVertex(id)
vertex
}
val triggerVertices: java.util.List[ExecutionJobVertex] =
snapshotSettings.getVerticesToTrigger().asScala.map(idToVertex).asJava
val ackVertices: java.util.List[ExecutionJobVertex] =
snapshotSettings.getVerticesToAcknowledge().asScala.map(idToVertex).asJava
val confirmVertices: java.util.List[ExecutionJobVertex] =
snapshotSettings.getVerticesToConfirm().asScala.map(idToVertex).asJava
val completedCheckpoints = checkpointRecoveryFactory
.createCompletedCheckpoints(jobId, userCodeLoader)
val checkpointIdCounter = checkpointRecoveryFactory.createCheckpointIDCounter(jobId)
executionGraph.enableSnapshotCheckpointing(
snapshotSettings.getCheckpointInterval,
snapshotSettings.getCheckpointTimeout,
snapshotSettings.getMinPauseBetweenCheckpoints,
snapshotSettings.getMaxConcurrentCheckpoints,
triggerVertices,
ackVertices,
confirmVertices,
context.system,
leaderSessionID.orNull,
checkpointIdCounter,
completedCheckpoints,
recoveryMode,
savepointStore)
}

JobManager自身会注册Job状态变更的事件回调:

1
executionGraph.registerJobStatusListener(new AkkaActorGateway(self, leaderSessionID.orNull))

如果Client也需要感知到执行结果以及Job状态的变更,那么也会为Client注册事件回调:

1
2
3
4
5
if (jobInfo.listeningBehaviour == ListeningBehaviour.EXECUTION_RESULT_AND_STATE_CHANGES) {
val gateway = new AkkaActorGateway(jobInfo.client, leaderSessionID.orNull)
executionGraph.registerExecutionListener(gateway)
executionGraph.registerJobStatusListener(gateway)
}

以上这些代码从将Job相关的Jar加入到类库缓存管理器开始,都被包裹在try块中,如果产生异常将进入catch代码块中进行异常处理:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
catch {
case t: Throwable =>
log.error(s"Failed to submit job $jobId ($jobName)", t)
libraryCacheManager.unregisterJob(jobId)
currentJobs.remove(jobId)
if (executionGraph != null) {
executionGraph.fail(t)
}
val rt: Throwable = if (t.isInstanceOf[JobExecutionException]) {
t
} else {
new JobExecutionException(jobId, s"Failed to submit job $jobId ($jobName)", t)
}
jobInfo.client ! decorateMessage(JobResultFailure(new SerializedThrowable(rt)))
return
}

异常处理时首先根据jobID移除类库缓存中跟当前Job有关的类库,接着从currentJobsMap中移除job对应的ExecutionGraph,JobInfo元组信息。然后调用ExecutionGraph的fail方法,促使其失败。最后,将产生的异常以JobResultFailure消息告知客户端并结束方法调用。

从当前开始直到最后的这段代码可能会造成阻塞,将会被包裹在future块中并以异步的方式执行。先判断当前的是否是恢复模式,如果是恢复模式则从最近的检查点恢复:

1
2
3
if (isRecovery) {
executionGraph.restoreLatestCheckpointedState()
}

如果不是恢复模式,但快照配置中存在保存点路径,也将基于保存点来重置状态:

1
executionGraph.restoreSavepoint(savepointPath)

然后会把当前的JobGraph信息写入SubmittedJobGraphStore,它主要用于恢复的目的

1
submittedJobGraphs.putJobGraph(new SubmittedJobGraph(jobGraph, jobInfo))

执行到这一步,就可以向Client回复JobSubmitSuccess消息了:

1
jobInfo.client ! decorateMessage(JobSubmitSuccess(jobGraph.getJobID))

接下来会基于ExecutionGraph触发Job的调度,这是Task被执行的前提:

1
2
3
4
5
if (leaderElectionService.hasLeadership) {
executionGraph.scheduleForExecution(scheduler)
} else {
self ! decorateMessage(RemoveJob(jobId, removeJobFromStateBackend = false))
}

为了防止多个JobManager同时调度相同的Job的情况产生,这里首先判断当前节点是否是Leader。如果是,才会进行调度。否则将会向自身发送一条RemoveJob消息,以进入其他处理逻辑。

到此为止,submitJob方法的梳理就算完成了。因为这是JobManager接收到Client提交的Job后的主要处理方法,所以包含的逻辑比较多。


微信扫码关注公众号:Apache_Flink

apache_flink_weichat


QQ扫码关注QQ群:Apache Flink学习交流群(123414680)

qrcode_for_apache_flink_qq_group