Flink Heartbeat of TaskManager和Heartbeat of ResourceManager timed out问题
最近上了个Flink任务,运行一段时间后就自动停止了,很是郁闷,查看最后一个chekpoint时间点,翻看时间日志
2019-12-13 07:25:24.566 flink [flink-akka.actor.default-dispatcher-41] INFO org.apache.flink.runtime.executiongraph.ExecutionGraph - Job PayOrder (88c9cc0c85875332cc5e4ed6418cd667) switched from state RUNNING to FAILING.java.util.concurrent.TimeoutException: Heartbeat of TaskManager with id container_1566481621886_4397244_01_000004 timed out.
at org.apache.flink.runtime.jobmaster.JobMaster$TaskManagerHeartbeatListener.notifyHeartbeatTimeout(JobMaster.java:1656)
at org.apache.flink.runtime.heartbeat.HeartbeatManagerImpl$HeartbeatMonitor.run(HeartbeatManagerImpl.java:339)
at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
at java.util.concurrent.FutureTask.run(FutureTask.java:266)
at org.apache.flink.runtime.concurrent.akka.ActorSystemScheduledExecutorAdapter$ScheduledFutureTask.run(ActorSystemScheduledExecutorAdapter.java:154)
at akka.dispatch.TaskInvocation.run(AbstractDispatcher.scala:39)
at akka.dispatch.ForkJoinExecutorConfigurator$AkkaForkJoinTask.exec(AbstractDispatcher.scala:415)
at scala.concurrent.forkjoin.ForkJoinTask.doExec(ForkJoinTask.java:260)
at scala.concurrent.forkjoin.ForkJoinPool$WorkQueue.runTask(ForkJoinPool.java:1339)
at scala.concurrent.forkjoin.ForkJoinPool.runWorker(ForkJoinPool.java:1979)
at scala.concurrent.forkjoin.ForkJoinWorkerThread.run(ForkJoinWorkerThread.java:107)
2019-12-13 07:25:24.519 flink [flink-akka.actor.default-dispatcher-20] INFO org.apache.flink.runtime.jobmaster.JobMaster - Close ResourceManager connection 9b7931812dbed76060b48a696d72a869: The heartbeat of ResourceManager with id 9b7931812dbed76060b48a696d72a869 timed out..
根据Heartbeat of TaskManager with id和The heartbeat of ResourceManager with id在源码中找出这样的代码
private class TaskManagerHeartbeatListener implements HeartbeatListener<AccumulatorReport, Void> {
private final JobMasterGateway jobMasterGateway;
private TaskManagerHeartbeatListener(JobMasterGateway jobMasterGateway) {
this.jobMasterGateway = Preconditions.checkNotNull(jobMasterGateway);
}
@Override
public void notifyHeartbeatTimeout(ResourceID resourceID) {
jobMasterGateway.disconnectTaskManager(
resourceID,
new TimeoutException("Heartbeat of TaskManager with id " + resourceID + " timed out."));
}
@Override
public void reportPayload(ResourceID resourceID, AccumulatorReport payload) {
for (AccumulatorSnapshot snapshot : payload.getAccumulatorSnapshots()) {
schedulerNG.updateAccumulators(snapshot);
}
}
@Override
public CompletableFuture<Void> retrievePayload(ResourceID resourceID) {
return CompletableFuture.completedFuture(null);
}
}
private class ResourceManagerHeartbeatListener implements HeartbeatListener<Void, Void> {
@Override
public void notifyHeartbeatTimeout(final ResourceID resourceId) {
runAsync(() -> {
log.info("The heartbeat of ResourceManager with id {} timed out.", resourceId);
if (establishedResourceManagerConnection != null && establishedResourceManagerConnection.getResourceManagerResourceID().equals(resourceId)) {
reconnectToResourceManager(
new JobMasterException(
String.format("The heartbeat of ResourceManager with id %s timed out.", resourceId)));
}
});
}
@Override
public void reportPayload(ResourceID resourceID, Void payload) {
// nothing to do since the payload is of type Void
}
@Override
public CompletableFuture<Void> retrievePayload(ResourceID resourceID) {
return CompletableFuture.completedFuture(null);
}
}
然后在这实例化
this.taskManagerHeartbeatManager = heartbeatServices.createHeartbeatManagerSender(resourceId,new TaskManagerHeartbeatListener(selfGateway),rpcService.getScheduledExecutor(),log);
顺着去heartbeatServices瞅瞅了
/**
* HeartbeatServices gives access to all services needed for heartbeating. This includes the
* creation of heartbeat receivers and heartbeat senders.
*/
public class HeartbeatServices { /** Heartbeat interval for the created services. */
protected final long heartbeatInterval; /** Heartbeat timeout for the created services. */
protected final long heartbeatTimeout; public HeartbeatServices(long heartbeatInterval, long heartbeatTimeout) {
Preconditions.checkArgument(0L < heartbeatInterval, "The heartbeat interval must be larger than 0.");
Preconditions.checkArgument(heartbeatInterval <= heartbeatTimeout, "The heartbeat timeout should be larger or equal than the heartbeat interval."); this.heartbeatInterval = heartbeatInterval;
this.heartbeatTimeout = heartbeatTimeout;
} /**
* Creates a heartbeat manager which does not actively send heartbeats.
*
* @param resourceId Resource Id which identifies the owner of the heartbeat manager
* @param heartbeatListener Listener which will be notified upon heartbeat timeouts for registered
* targets
* @param scheduledExecutor Scheduled executor to be used for scheduling heartbeat timeouts
* @param log Logger to be used for the logging
* @param <I> Type of the incoming payload
* @param <O> Type of the outgoing payload
* @return A new HeartbeatManager instance
*/
public <I, O> HeartbeatManager<I, O> createHeartbeatManager(
ResourceID resourceId,
HeartbeatListener<I, O> heartbeatListener,
ScheduledExecutor scheduledExecutor,
Logger log) { return new HeartbeatManagerImpl<>(
heartbeatTimeout,
resourceId,
heartbeatListener,
scheduledExecutor,
scheduledExecutor,
log);
} /**
* Creates a heartbeat manager which actively sends heartbeats to monitoring targets.
*
* @param resourceId Resource Id which identifies the owner of the heartbeat manager
* @param heartbeatListener Listener which will be notified upon heartbeat timeouts for registered
* targets
* @param scheduledExecutor Scheduled executor to be used for scheduling heartbeat timeouts
* @param log Logger to be used for the logging
* @param <I> Type of the incoming payload
* @param <O> Type of the outgoing payload
* @return A new HeartbeatManager instance which actively sends heartbeats
*/
public <I, O> HeartbeatManager<I, O> createHeartbeatManagerSender(
ResourceID resourceId,
HeartbeatListener<I, O> heartbeatListener,
ScheduledExecutor scheduledExecutor,
Logger log) { return new HeartbeatManagerSenderImpl<>(
heartbeatInterval,
heartbeatTimeout,
resourceId,
heartbeatListener,
scheduledExecutor,
scheduledExecutor,
log);
} /**
* Creates an HeartbeatServices instance from a {@link Configuration}.
*
* @param configuration Configuration to be used for the HeartbeatServices creation
* @return An HeartbeatServices instance created from the given configuration
*/
public static HeartbeatServices fromConfiguration(Configuration configuration) {
long heartbeatInterval = configuration.getLong(HeartbeatManagerOptions.HEARTBEAT_INTERVAL); long heartbeatTimeout = configuration.getLong(HeartbeatManagerOptions.HEARTBEAT_TIMEOUT); return new HeartbeatServices(heartbeatInterval, heartbeatTimeout);
}
}
没错超时时间就在HeartbeatManagerOptions.HEARTBEAT_TIMEOUT
/** Timeout for requesting and receiving heartbeat for both sender and receiver sides. */
public static final ConfigOption<Long> HEARTBEAT_TIMEOUT =
key("heartbeat.timeout")
.defaultValue(50000L)
.withDescription("Timeout for requesting and receiving heartbeat for both sender and receiver sides.");
引起心跳超时有可能是yarn压力比较大引起的,先暂时在conf/flink-conf.yaml将这个值调大一点,再观察。
#Timeout for requesting and receiving heartbeat for both sender and receiver sides.
heartbeat.timeout: 180000

Flink Heartbeat of TaskManager和Heartbeat of ResourceManager timed out问题的更多相关文章
- Flink JobManager 和 TaskManager 原理
转自:https://www.cnblogs.com/nicekk/p/11561836.html 一.概述 Flink 整个系统主要由两个组件组成,分别为 JobManager 和 TaskMana ...
- Linux-HA实战(1)— Heartbeat安装
接触Heartbeat主要是因为之前项目中使用了TFS,最近想给nameserver做HA,因为TFS官方用的Heartbeat,所以刚好了解下,参考了网络上很多内容,这里简单记录下. 内容 环境和软 ...
- 高可用集群heartbeat全攻略
heartbeat的概念 Linux-HA的全称是High-Availability Linux,它是一个开源项目,这个开源项目的目标是:通过社区开发者的共同努力,提供一个增强linux可靠性(r ...
- 使用Heartbeat实现双机热备
使用Heartbeat实现"双机热备"或者称为"双机互备"heartbeat的工作原理:heartbeat最核心的包含两个部分,心跳监測部分和资源接管部分,心跳 ...
- Heartbeat实现集群高可用热备
公司最近需要针对服务器实现热可用热备,这几天也一直在琢磨这个方面的东西,今天做了一些Heartbeat方面的工作,在此记录下来,给需要的人以参考. Heartbeat 项目是 Linux-HA 工程的 ...
- Heartbeat详解
转自:http://blog.sina.com.cn/s/blog_7b6fc4c901012om0.html 配置主节点的Heartbeat Heartbeat的主要配置文件有ha.cf.hares ...
- (转)Linux-HA实战(1)— Heartbeat安装
原文:http://blog.csdn.net/liaomin416100569/article/details/76087448-------centos7源代码编译安装heartbeat 原文:h ...
- Heartbeat使用梳理
在日常的集群系统架构中,一般用到Heartbeat的主要就2种:1)高可用(High Availability)HA集群, 使用Heartbeat实现,也称为"双机热备", &qu ...
- 1 NFS高可用解决方案之DRBD+heartbeat搭建
preface NFS作为业界常用的共享存储方案,被众多公司采用.我司也不列外,使用NFS作为共享存储,为前端WEB server提供服务,主要存储网页代码以及其他文件. 高可用方案 说道NFS,不得 ...
- HA(High available)--Heartbeat高可用性集群(双机热备)菜鸟入门级
HA(High available)--Heartbeat高可用性集群(双机热备) 1.理解:两台服务器A和B ,当A提供服务,B闲置待命,当A服务宕机,会自动切换至B机器继续提供服务.当主机恢复 ...
随机推荐
- 合格できる日本語能力試験, N1.PDF
书本详情 合格できる日本語能力試験, N1种类:Languages - Japanese Language Reference年:2010出版:Shohan.出版社:Aruku语言:japanese页 ...
- js,php中的面向对象
面向对象 js中的面向对象: 面向对象(OOP):通过函数封装得到的一个类 1.类的本质还是函数 每个类(函数)天生有一个prototype的属性,这个prototype又是一个对象,这个对象里有个c ...
- signalR从外部服务中推送消息和全局参数的设置
在前面的章节中,我们都是采用两方模式, 客户端 <-> 服务端,也就是说在这种情况下如果有第三方如果有通知的没有办法插足的,下面解决方案 1:和之前的代码一样,在index页面加一个客户端 ...
- linux 中后台运行python脚本
nohup python yourscript.py &可以让你的程序在后台运行,控制台输出导向nohup.out文件 使用nobup命令 结尾处加一个& 符号
- Vue.js的使用经验
Vue.js的使用经验 Vue.js的意义 解耦了视图与数据 可复用的组件 前端路由 状态管理 虚拟DOM Vue提供了很多实例属性与方法,都以$开头 $el可以访问挂载Vue实例的元素. $se ...
- DSP(数字信号处理)与监控摄像机相关技术解读
提起安防领域,我们第一个想到的应该就是监控摄像机了,它对于安防的重要性,就像人的眼睛对于人的作用一样,是重要的信息获取来源. 监控摄像机广泛应用于学校.公司.银行.交通.平安城市等多个安保领域. 广州 ...
- 7.mysql索引失效
失效的七字口诀: 模型数空运最快 模:模糊查询以%开始索引失效: 型:数据类型转换 函数:函数的索引 空:索引列为空不走索引, 运:对索引列进行加减乘除会失效 最:不按聚合索引的最左匹配会 ...
- easyExcel-modle
package com.alibaba.easyexcel.test.model;import com.alibaba.excel.annotation.ExcelProperty;import co ...
- shell脚本,shell语法和结构(以Cshell/TC shell为例)
1.C shell/TC shell常用语法和结构 (1) shbang行: #!/bin/sh,通知内核使用哪种内核解释脚本; #!/bin/csh 或 #!/bin/tcsh (2) 注释: 以# ...
- uniapp记录
1.uniapp获取imei imsi 型号 厂商 https://blog.csdn.net/qq_35620498/article/details/112172115