观察了下。CPU什么的有尖刺，但是也算基本正常，因为我的任务就是5分钟一波。基本每5分钟都有个尖刺。 然后目前通过Flink的web-ui看了下gc情况。 发现部分集群的fgc的确有问题，fgc平均大概达到10-20s，当然只有平均值，不清楚是否有某些gc时间更长情况。总难题来说10-20s的确是比较长的，这个我之后会去看看改进下。
（1）不过不清楚这个是否和这个问题直接相关，因为20s的卡顿是否足以引起该问题呢？ （2）此外，大家推荐个内存设置，比如你们都多少TM，每个TM多少内存，跑的任务多大数据量大概。 我目前5个TM的集群，单TM100G内存，跑任务大概10w qps的入口流量，但是很大部分呢会过滤掉，后续部分流量较少。此外，检查点大概达到3-4GB。 Michael Ran <[email protected]> 于2021年3月9日周二 下午4:27写道： > 看看当时的负载呢？有没有过高的情况，是什么原因。然后监控下网络和磁盘 > 在 2021-03-09 14:57:43，"yidan zhao" <[email protected]> 写道： > >而且大家推荐怎么设置呢，我可能默认就G1了。不清楚G1是否也需要精调。 > >我目前设置的内存还是比较大的。（50G的，100G的TaskManager都有），这么大heap，是否需要特别设置啥呢？ > > > >或者是否有必要拆小，比如设置10Gheap，然后把taskmanager数量提上去。 > > > >yidan zhao <[email protected]> 于2021年3月9日周二 下午2:56写道： > > > >> 好的，我会看下。 > >> 然后我今天发现我好多个集群GC collector不一样。 > >> 目前发现3种，默认的是G1。flink conf中配置了env.java.opts: > >> "-XX:-OmitStackTraceInFastThrow"的情况出现了2种，一种是Parallel GC with 83 > >> threads，还有一种是Mark Sweep Compact GC。 > >> 大佬们，Flink是根据内存大小有什么动态调整吗。 > >> > >> > >> > 不使用G1我大概理解了，可能设置了java.opts这个是覆盖，不是追加。本身我只是希望设置下-XX:-OmitStackTraceInFastThrow而已。 > >> > >> > >> 杨杰 <[email protected]> 于2021年3月8日周一 下午3:09写道： > >> > >>> Hi, > >>> > >>> 可以排查下 GC 情况，频繁 FGC 也会导致这些情况。 > >>> > >>> Best, > >>> jjiey > >>> > >>> > 2021年3月8日 14:37，yidan zhao <[email protected]> 写道： > >>> > > >>> > > >>> > 如题，我有个任务频繁发生该异常然后重启。今天任务启动1h后，看了下WEB-UI的检查点也没，restored达到了8已经。然后Exception页面显示该错误，估计大多数都是因为该错误导致的restore。 > >>> > 除此外，就是 ‘Job leader for job id eb5d2893c4c6f4034995b9c8e180f01e lost > >>> > leadership’ 错导致任务重启。 > >>> > > >>> > 下面给出刚刚的一个错误日志（环境flink1.12，standalone集群，5JM+5TM，JM和TM混部在相同机器）： > >>> > 2021-03-08 14:31:40 > >>> > org.apache.flink.runtime.io > >>> .network.netty.exception.RemoteTransportException: > >>> > Error at remote task manager '10.35.185.38/10.35.185.38:2016'. > >>> > at org.apache.flink.runtime.io.network.netty. > >>> > CreditBasedPartitionRequestClientHandler.decodeMsg( > >>> > CreditBasedPartitionRequestClientHandler.java:294) > >>> > at org.apache.flink.runtime.io.network.netty. > >>> > CreditBasedPartitionRequestClientHandler.channelRead( > >>> > CreditBasedPartitionRequestClientHandler.java:183) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > AbstractChannelHandlerContext.invokeChannelRead( > >>> > AbstractChannelHandlerContext.java:379) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > AbstractChannelHandlerContext.invokeChannelRead( > >>> > AbstractChannelHandlerContext.java:365) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > > >>> > AbstractChannelHandlerContext.fireChannelRead(AbstractChannelHandlerContext > >>> > .java:357) > >>> > at org.apache.flink.runtime.io.network.netty. > >>> > NettyMessageClientDecoderDelegate.channelRead( > >>> > NettyMessageClientDecoderDelegate.java:115) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > AbstractChannelHandlerContext.invokeChannelRead( > >>> > AbstractChannelHandlerContext.java:379) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > AbstractChannelHandlerContext.invokeChannelRead( > >>> > AbstractChannelHandlerContext.java:365) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > > >>> > AbstractChannelHandlerContext.fireChannelRead(AbstractChannelHandlerContext > >>> > .java:357) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > > >>> > DefaultChannelPipeline$HeadContext.channelRead(DefaultChannelPipeline.java: > >>> > 1410) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > AbstractChannelHandlerContext.invokeChannelRead( > >>> > AbstractChannelHandlerContext.java:379) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > AbstractChannelHandlerContext.invokeChannelRead( > >>> > AbstractChannelHandlerContext.java:365) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > > DefaultChannelPipeline.fireChannelRead(DefaultChannelPipeline.java:919) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel.epoll. > >>> > AbstractEpollStreamChannel$EpollStreamUnsafe.epollInReady( > >>> > AbstractEpollStreamChannel.java:792) > >>> > at > >>> org.apache.flink.shaded.netty4.io.netty.channel.epoll.EpollEventLoop > >>> > .processReady(EpollEventLoop.java:475) > >>> > at > >>> org.apache.flink.shaded.netty4.io.netty.channel.epoll.EpollEventLoop > >>> > .run(EpollEventLoop.java:378) > >>> > at org.apache.flink.shaded.netty4.io.netty.util.concurrent. > >>> > SingleThreadEventExecutor$4.run(SingleThreadEventExecutor.java:989) > >>> > at org.apache.flink.shaded.netty4.io.netty.util.internal. > >>> > ThreadExecutorMap$2.run(ThreadExecutorMap.java:74) > >>> > at java.lang.Thread.run(Thread.java:748) > >>> > Caused by: org.apache.flink.runtime.io.network.partition. > >>> > ProducerFailedException: org.apache.flink.util.FlinkException: > >>> JobManager > >>> > responsible for eb5d2893c4c6f4034995b9c8e180f01e lost the leadership. > >>> > at org.apache.flink.runtime.io > .network.netty.PartitionRequestQueue > >>> > .writeAndFlushNextMessageIfPossible(PartitionRequestQueue.java:221) > >>> > at org.apache.flink.runtime.io > .network.netty.PartitionRequestQueue > >>> > .enqueueAvailableReader(PartitionRequestQueue.java:108) > >>> > at org.apache.flink.runtime.io > .network.netty.PartitionRequestQueue > >>> > .userEventTriggered(PartitionRequestQueue.java:170) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > AbstractChannelHandlerContext.invokeUserEventTriggered( > >>> > AbstractChannelHandlerContext.java:346) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > AbstractChannelHandlerContext.invokeUserEventTriggered( > >>> > AbstractChannelHandlerContext.java:332) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > AbstractChannelHandlerContext.fireUserEventTriggered( > >>> > AbstractChannelHandlerContext.java:324) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > > >>> > ChannelInboundHandlerAdapter.userEventTriggered(ChannelInboundHandlerAdapter > >>> > .java:117) > >>> > at org.apache.flink.shaded.netty4.io.netty.handler.codec. > >>> > > ByteToMessageDecoder.userEventTriggered(ByteToMessageDecoder.java:365) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > AbstractChannelHandlerContext.invokeUserEventTriggered( > >>> > AbstractChannelHandlerContext.java:346) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > AbstractChannelHandlerContext.invokeUserEventTriggered( > >>> > AbstractChannelHandlerContext.java:332) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > AbstractChannelHandlerContext.fireUserEventTriggered( > >>> > AbstractChannelHandlerContext.java:324) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > > >>> > DefaultChannelPipeline$HeadContext.userEventTriggered(DefaultChannelPipeline > >>> > .java:1428) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > AbstractChannelHandlerContext.invokeUserEventTriggered( > >>> > AbstractChannelHandlerContext.java:346) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > AbstractChannelHandlerContext.invokeUserEventTriggered( > >>> > AbstractChannelHandlerContext.java:332) > >>> > at org.apache.flink.shaded.netty4.io.netty.channel. > >>> > > >>> > DefaultChannelPipeline.fireUserEventTriggered(DefaultChannelPipeline.java: > >>> > 913) > >>> > at org.apache.flink.runtime.io > .network.netty.PartitionRequestQueue > >>> > .lambda$notifyReaderNonEmpty$0(PartitionRequestQueue.java:87) > >>> > at org.apache.flink.shaded.netty4.io.netty.util.concurrent. > >>> > AbstractEventExecutor.safeExecute(AbstractEventExecutor.java:164) > >>> > at org.apache.flink.shaded.netty4.io.netty.util.concurrent. > >>> > > >>> > SingleThreadEventExecutor.runAllTasks(SingleThreadEventExecutor.java:472) > >>> > at > >>> org.apache.flink.shaded.netty4.io.netty.channel.epoll.EpollEventLoop > >>> > .run(EpollEventLoop.java:387) > >>> > ... 3 more > >>> > Caused by: org.apache.flink.util.FlinkException: JobManager > responsible > >>> for > >>> > eb5d2893c4c6f4034995b9c8e180f01e lost the leadership. > >>> > at org.apache.flink.runtime.taskexecutor.TaskExecutor > >>> > .disconnectJobManagerConnection(TaskExecutor.java:1422) > >>> > at org.apache.flink.runtime.taskexecutor.TaskExecutor.access$1300( > >>> > TaskExecutor.java:174) > >>> > at org.apache.flink.runtime.taskexecutor. > >>> > > TaskExecutor$JobLeaderListenerImpl.lambda$null$2(TaskExecutor.java:1856) > >>> > at java.util.Optional.ifPresent(Optional.java:159) > >>> > at org.apache.flink.runtime.taskexecutor. > >>> > TaskExecutor$JobLeaderListenerImpl.lambda$jobManagerLostLeadership$3( > >>> > TaskExecutor.java:1855) > >>> > at org.apache.flink.runtime.rpc.akka.AkkaRpcActor.handleRunAsync( > >>> > AkkaRpcActor.java:404) > >>> > at > org.apache.flink.runtime.rpc.akka.AkkaRpcActor.handleRpcMessage( > >>> > AkkaRpcActor.java:197) > >>> > at org.apache.flink.runtime.rpc.akka.AkkaRpcActor.handleMessage( > >>> > AkkaRpcActor.java:154) > >>> > at akka.japi.pf.UnitCaseStatement.apply(CaseStatements.scala:26) > >>> > at akka.japi.pf.UnitCaseStatement.apply(CaseStatements.scala:21) > >>> > at > scala.PartialFunction$class.applyOrElse(PartialFunction.scala:123) > >>> > at akka.japi.pf > >>> .UnitCaseStatement.applyOrElse(CaseStatements.scala:21) > >>> > at > >>> scala.PartialFunction$OrElse.applyOrElse(PartialFunction.scala:170) > >>> > at > >>> scala.PartialFunction$OrElse.applyOrElse(PartialFunction.scala:171) > >>> > at > >>> scala.PartialFunction$OrElse.applyOrElse(PartialFunction.scala:171) > >>> > at akka.actor.Actor$class.aroundReceive(Actor.scala:517) > >>> > at akka.actor.AbstractActor.aroundReceive(AbstractActor.scala:225) > >>> > at akka.actor.ActorCell.receiveMessage(ActorCell.scala:592) > >>> > at akka.actor.ActorCell.invoke(ActorCell.scala:561) > >>> > at akka.dispatch.Mailbox.processMailbox(Mailbox.scala:258) > >>> > at akka.dispatch.Mailbox.run(Mailbox.scala:225) > >>> > at akka.dispatch.Mailbox.exec(Mailbox.scala:235) > >>> > at > akka.dispatch.forkjoin.ForkJoinTask.doExec(ForkJoinTask.java:260) > >>> > at > akka.dispatch.forkjoin.ForkJoinPool$WorkQueue.runTask(ForkJoinPool > >>> > .java:1339) > >>> > at > >>> akka.dispatch.forkjoin.ForkJoinPool.runWorker(ForkJoinPool.java:1979) > >>> > at > >>> akka.dispatch.forkjoin.ForkJoinWorkerThread.run(ForkJoinWorkerThread > >>> > .java:107) > >>> > Caused by: java.lang.Exception: Job leader for job id > >>> > eb5d2893c4c6f4034995b9c8e180f01e lost leadership. > >>> > ... 24 more > >>> > > >>> > > >>> > （1）zookeeper的超时设置的是60s，感觉网络异常zk超时不至于60s都不够。 > >>> > （2）akka.ask.timeout: 60s > >>> > taskmanager.network.request-backoff.max： 60000 > >>> > akka此参数之前也调整为60s了。 > >>> > > >>> > 如上信息，希望社区同学们给点思路。 > >>> > > >>> > >>> >
