在XD中部署了修改过的支持Yarn Client HA的Spring Hadoop Yarn版本,运行一段时间后出现异常:
2017-03-31 14:52:01,243 DEBUG [DefaultContainerLauncher] - Checking status of containers previousely launched
2017-03-31 14:52:02,329 WARN [Client$Connection$1] - Exception encountered while connecting to the server :
org.apache.hadoop.ipc.RemoteException(org.apache.hadoop.security.token.SecretManager$InvalidToken): Invalid AMRMToken from appattempt_1490607357271_0001_000002
2017-03-31 14:52:02,331 WARN [RetryInvocationHandler] - Exception while invoking class org.apache.hadoop.yarn.api.impl.pb.client.ApplicationMasterProtocolPBClientImpl.allocate over rm2. Not retrying because failovers (30) exceeded maximum allowed (30) org.apache.hadoop.security.token.SecretManager$InvalidToken: Invalid AMRMToken from appattempt_1490607357271_0001_000002
at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)
at sun.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:62)
at sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45)
at java.lang.reflect.Constructor.newInstance(Constructor.java:422) at org.apache.hadoop.yarn.ipc.RPCUtil.instantiateException(RPCUtil.java:53)
at org.apache.hadoop.yarn.ipc.RPCUtil.unwrapAndThrowException(RPCUtil.java:104)
at org.apache.hadoop.yarn.api.impl.pb.client.ApplicationMasterProtocolPBClientImpl.allocate(ApplicationMasterProtocolPBClientImpl.java:79)
调查RM的日志发现在该时间段,出现RM与ZK无法通信的情况,导致主备RM均进入neutral state,因此AM与RM无法通信。
RM恢复后,由于长时间没有收到AM的心跳,因此将AM kill掉。
根据默认重试机制,会重启一次AM。第二次则不再重启。
相关配置项:
yarn.am.liveness-monitor.expiry-interval-ms,默认10min
因此,在网络不稳的环境下,建议将上述配置项时间延长,以给RM恢复AM的时机。