[YARN-4431] Not necessary to do unRegisterNM() if NM get stop due to failed to connect to RM

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 2.8.0, 3.0.0-alpha1
• Component/s: nodemanager
• Hadoop Flags: Reviewed
• Assignee: Junping Du
• Reporter: Junping Du
• Created: 07/Dec/15 18:31
• Updated: 25/Oct/19 20:25
• Resolved: 09/Dec/15 05:39

Description

2015-12-07 12:16:57,873 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: 0.0.0.0/0.0.0.0:8031. Already tried 8 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS)
2015-12-07 12:16:58,874 INFO org.apache.hadoop.ipc.Client: Retrying connect to server: 0.0.0.0/0.0.0.0:8031. Already tried 9 time(s); retry policy is RetryUpToMaximumCountWithFixedSleep(maxRetries=10, sleepTime=1000 MILLISECONDS)
2015-12-07 12:16:58,876 WARN org.apache.hadoop.yarn.server.nodemanager.NodeStatusUpdaterImpl: Unregistration of the Node 10.200.10.53:25454 failed.
java.net.ConnectException: Call From jduMBP.local/10.200.10.53 to 0.0.0.0:8031 failed on connection exception: java.net.ConnectException: Connection refused; For more details see:  http://wiki.apache.org/hadoop/ConnectionRefused
        at sun.reflect.GeneratedConstructorAccessor30.newInstance(Unknown Source)
        at sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45)
        at java.lang.reflect.Constructor.newInstance(Constructor.java:408)
        at org.apache.hadoop.net.NetUtils.wrapWithMessage(NetUtils.java:792)
        at org.apache.hadoop.net.NetUtils.wrapException(NetUtils.java:732)
        at org.apache.hadoop.ipc.Client.call(Client.java:1452)
        at org.apache.hadoop.ipc.Client.call(Client.java:1385)
        at org.apache.hadoop.ipc.ProtobufRpcEngine$Invoker.invoke(ProtobufRpcEngine.java:230)
        at com.sun.proxy.$Proxy74.unRegisterNodeManager(Unknown Source)
        at org.apache.hadoop.yarn.server.api.impl.pb.client.ResourceTrackerPBClientImpl.unRegisterNodeManager(ResourceTrackerPBClientImpl.java:98)
        at sun.reflect.GeneratedMethodAccessor13.invoke(Unknown Source)
        at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
        at java.lang.reflect.Method.invoke(Method.java:483)
        at org.apache.hadoop.io.retry.RetryInvocationHandler.invokeMethod(RetryInvocationHandler.java:255)
        at org.apache.hadoop.io.retry.RetryInvocationHandler.invoke(RetryInvocationHandler.java:103)
        at com.sun.proxy.$Proxy75.unRegisterNodeManager(Unknown Source)
        at org.apache.hadoop.yarn.server.nodemanager.NodeStatusUpdaterImpl.unRegisterNM(NodeStatusUpdaterImpl.java:267)
        at org.apache.hadoop.yarn.server.nodemanager.NodeStatusUpdaterImpl.serviceStop(NodeStatusUpdaterImpl.java:245)
        at org.apache.hadoop.service.AbstractService.stop(AbstractService.java:221)
        at org.apache.hadoop.service.ServiceOperations.stop(ServiceOperations.java:52)
        at org.apache.hadoop.service.ServiceOperations.stopQuietly(ServiceOperations.java:80)
        at org.apache.hadoop.service.CompositeService.stop(CompositeService.java:157)
        at org.apache.hadoop.service.CompositeService.serviceStop(CompositeService.java:131)
        at org.apache.hadoop.yarn.server.nodemanager.NodeManager.serviceStop(NodeManager.java:377)

If RM down for some reason, NM's NodeStatusUpdaterImpl will retry the connection with proper retry policy. After retry the maximum times (15 minutes by default), it will send NodeManagerEventType.SHUTDOWN to shutdown NM. But NM shutdown will call NodeStatusUpdaterImpl.serviceStop() which will call unRegisterNM() to unregister NM from RM and get retry again (another 15 minutes). This is completely unnecessary and we should skip unRegisterNM when NM get shutdown because of connection issues.