Standalone模式下，通过Systemd管理Flink1.11.1的启停及异常退出

Flink以Standalone模式运行时，可能会发生jobmanager（以下简称jm）或taskmanager（以下简称tm）异常退出的情况，我们可以使用Linux自带的Systemd方式管理jm以及tm的启停，并在jm或tm出现故障时，及时将jm以及tm拉起来。

Flink在1.11版本后，从发行版中移除了对Hadoop的依赖包，如果需要使用Hadoop的一些特性，有两种解决方案：

【注】以下假设java、flink、hadoop都安装在/opt目录下，并且都建立了软连接：

1.设置HADOOP_CLASSPATH环境变量（推荐方案）

在安装了Flink的所有节点上，在/etc/profile中进行如下设置：

# Hadoop Envexport HADOOP_HOME=/opt/hadoopexport PATH=$PATH:$HADOOP_HOME/bin:$HADOOP_HOME/sbinexport HADOOP_CLASSPATH=`hadoop classpath`

然后通过以下命令使环境变量生效

sudo source /etc/profile

2.下载flink-shaded-hadoop-2-uber对应的jar包，并拷贝到Flink安装路径的lib目录下

下载地址：https://flink.apache.org/downloads.html#additional-components

由于以systemd方式启动时，系统设置的环境变量，在.service文件中是不能使用的，所以需要在.service文件中单独显式设置环境变量：

1./usr/lib/systemd/system/flink-jobmanager.service

[Unit]
Description=Flink Job Manager
After=syslog.target network.target remote-fs.target nss-lookup.target network-online.target
Requires=network-online.target

[Service]
User=teld
Group=teld
Type=forking
Environment=PATH=/opt/java/bin:/opt/flink/bin:/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin:/root/bin
Environment=JAVA_HOME=/opt/java
Environment=FLINK_HOME=/opt/flink
Environment=HADOOP_CLASSPATH=/opt/hadoop/etc/hadoop:/opt/hadoop/share/hadoop/common/lib/*:/opt/hadoop/share/hadoop/common/*:/opt/hadoop/
share/hadoop/hdfs:/opt/hadoop/share/hadoop/hdfs/lib/*:/opt/hadoop/share/hadoop/hdfs/*:/opt/hadoop/share/hadoop/yarn/lib/*:/opt/hadoop/sh
are/hadoop/yarn/*:/opt/hadoop/share/hadoop/mapreduce/lib/*:/opt/hadoop/share/hadoop/mapreduce/*:/opt/hadoop/contrib/capacity-scheduler/*
.jar
ExecStart=/opt/flink/bin/jobmanager.sh start
ExecStop=/opt/flink/bin/jobmanager.sh stop

Restart=on-failure

[Install]
WantedBy=multi-user.target

【注】HADOOP_CLASSPATH对应的值，是通过执行以下命令获得到的：

hadoop classpath

2./usr/lib/systemd/system/flink-taskmanager.service

[Unit]
Description=Flink Task Manager
After=syslog.target network.target remote-fs.target nss-lookup.target network-online.target
Requires=network-online.target

[Service]
User=teld
Group=teld
Type=forking
Environment=PATH=/opt/java/bin:/opt/flink/bin:/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin:/root/bin
Environment=JAVA_HOME=/opt/java
Environment=FLINK_HOME=/opt/flink
Environment=HADOOP_CLASSPATH=/opt/hadoop/etc/hadoop:/opt/hadoop/share/hadoop/common/lib/*:/opt/hadoop/share/hadoop/common/*:/opt/hadoop/
share/hadoop/hdfs:/opt/hadoop/share/hadoop/hdfs/lib/*:/opt/hadoop/share/hadoop/hdfs/*:/opt/hadoop/share/hadoop/yarn/lib/*:/opt/hadoop/sh
are/hadoop/yarn/*:/opt/hadoop/share/hadoop/mapreduce/lib/*:/opt/hadoop/share/hadoop/mapreduce/*:/opt/hadoop/contrib/capacity-scheduler/*
.jar
ExecStart=/opt/flink/bin/taskmanager.sh start
ExecStop=/opt/flink/bin/taskmanager.sh stop

Restart=on-failure

[Install]
WantedBy=multi-user.target

【注】HADOOP_CLASSPATH对应的值，是通过执行以下命令获得到的：

hadoop classpath

通过sudo systemctl daemon-reload命令来加载上面针对jm以及tm的配置后，就可以使用Systemd的方式来管理jm以及tm了，并且能够在jm以及tm异常退出时，及时将它们拉起来：

sudo systemctl start flink-jobmanager.service

sudo systemctl stop flink-jobmanager.service

sudo systemctl status flink-jobmanager.service

sudo systemctl start flink-taskmanager.service

sudo systemctl stop flink-taskmanager.service

sudo systemctl status flink-jobmanager.service

遇到的坑：

1.如果Flink设置了启用Checkpoint，但是没有设置HADOOP_CLASSPATH环境变量，则提交job的时候，会报如下异常：

Caused by: org.apache.flink.util.FlinkRuntimeException: Failed to create checkpoint storage at checkpoint coordinator side.
at org.apache.flink.runtime.checkpoint.CheckpointCoordinator.<init>(CheckpointCoordinator.java:304)
at org.apache.flink.runtime.checkpoint.CheckpointCoordinator.<init>(CheckpointCoordinator.java:223)
at org.apache.flink.runtime.executiongraph.ExecutionGraph.enableCheckpointing(ExecutionGraph.java:483)
at org.apache.flink.runtime.executiongraph.ExecutionGraphBuilder.buildGraph(ExecutionGraphBuilder.java:338)
at org.apache.flink.runtime.scheduler.SchedulerBase.createExecutionGraph(SchedulerBase.java:269)
at org.apache.flink.runtime.scheduler.SchedulerBase.createAndRestoreExecutionGraph(SchedulerBase.java:242)
at org.apache.flink.runtime.scheduler.SchedulerBase.<init>(SchedulerBase.java:229)
at org.apache.flink.runtime.scheduler.DefaultScheduler.<init>(DefaultScheduler.java:119)
at org.apache.flink.runtime.scheduler.DefaultSchedulerFactory.createInstance(DefaultSchedulerFactory.java:103)
at org.apache.flink.runtime.jobmaster.JobMaster.createScheduler(JobMaster.java:284)
at org.apache.flink.runtime.jobmaster.JobMaster.<init>(JobMaster.java:272)
at org.apache.flink.runtime.jobmaster.factories.DefaultJobMasterServiceFactory.createJobMasterService(DefaultJobMasterServiceFac
tory.java:98)
at org.apache.flink.runtime.jobmaster.factories.DefaultJobMasterServiceFactory.createJobMasterService(DefaultJobMasterServiceFac
tory.java:40)
at org.apache.flink.runtime.jobmaster.JobManagerRunnerImpl.<init>(JobManagerRunnerImpl.java:140)
at org.apache.flink.runtime.dispatcher.DefaultJobManagerRunnerFactory.createJobManagerRunner(DefaultJobManagerRunnerFactory.java
:84)
at org.apache.flink.runtime.dispatcher.Dispatcher.lambda$createJobManagerRunner$6(Dispatcher.java:388)
... 7 more
Caused by: org.apache.flink.core.fs.UnsupportedFileSystemSchemeException: Could not find a file system implementation for scheme 'hdfs'.
 The scheme is not directly supported by Flink and no Hadoop file system to support this scheme could be loaded. For a full list of supp

2.在为flink-jobmanager.service以及flink-taskmanager.service中的HADOOP_CLASSPATH环境变量赋值时，尝试使用过反引号，期望将反引号内的Linux命令执行结果赋予变量，但实际上并不会执行反引号中的内容：

Environment=HADOOP_CLASSPATH=`/opt/hadoop/bin/hadoop classpath`

最后只得将直接执行hadoop classpath获得的结果，粘贴到.service文件中

Environment=HADOOP_CLASSPATH=/opt/hadoop/etc/hadoop:/opt/hadoop/share/hadoop/common/lib/*:/opt/hadoop/share/hadoop/common/*:/opt/hadoop/
share/hadoop/hdfs:/opt/hadoop/share/hadoop/hdfs/lib/*:/opt/hadoop/share/hadoop/hdfs/*:/opt/hadoop/share/hadoop/yarn/lib/*:/opt/hadoop/sh
are/hadoop/yarn/*:/opt/hadoop/share/hadoop/mapreduce/lib/*:/opt/hadoop/share/hadoop/mapreduce/*:/opt/hadoop/contrib/capacity-scheduler/*
.jar