You are viewing a plain text version of this content. The canonical link for it is here.

Posted to user@storm.apache.org by Jade <jw...@naver.com> on 2017/12/01 04:34:32 UTC

In storm 1.0.4, Worker died after kafka zookeeper process down

Hello.

I'm trying to test clustering storm 1.0.4 and storm-kafka.
(3 kafka(broker3+zookeeper3), and 3 storm cluster(zookeeper3+nimbus#+supervisor3))
 
Tested 3 case. But i saw some symptoms.
 
 
1. Shutdown kafka broker after deploing topology
-&gt; Going well. Occur error java.nio.channels.ClosedChannelException. And read partition after refreshing partition manager connections.
 
2. Shutdown zookeeper(in kafka server) after deploy topology
-&gt; Occur Connection Loss Error and finaly worker died.
Exception: Caused by: java.lang.RuntimeException: java.lang.RuntimeException: org.apache.zookeeper.KeeperException$ConnectionLossException: KeeperErrorCode = ConnectionLoss for /brokers/topics/xxxxx
                 ERROR o.a.s.util - Halting process: ("Worker died") java.lang.RuntimeException: ("Worker died")
 
3. Shutdown zookeeper(in storm server)

-&gt; Down One of zookeeper cluster, killed all storm process(nimbus, supervisor)
 

In No2, 3 case, Can available keep topology and storm going well except process down(or restart)?