Seems like that is due to zk connections from your hbase code, not from flume.
Thanks, Hari On Fri, Oct 31, 2014 at 4:30 PM, terreyshih <[email protected]> wrote: > There is no full stack trace. It seems to be come from the thread in > o.a.z.ClientCnxn.run. > In our logging scheme, we use the first initial of the fully qualified class > path, so the o.a.z.ClientCnxn.run would refer to > org.apache.zookeeper.ClientCnxn > Having said, that, it might not be flume related issue. The sink actually is > a customized sink that does some business logic and writes the output to > HBase. So this could be that some issues with Hbase cluster connection ? > any comments would be appreciated. > thx > 2014-10-31 15:31:06,790 DEBUG > [SinkRunner-PollingRunner-DefaultSinkProcessor-SendThread(worker-n3.sc.mynetworks.com:2181)] > [o.a.z.ClientCnxn.run:1171] - An exception was thrown while closing send > thread for session 0x448485df39f836c : Unable to read additional data from > server sessionid 0x448485df39f836c, likely server has closed socket >> On Oct 31, 2014, at 4:18 PM, Hari Shreedharan <[email protected]> >> wrote: >> >> Can you send the full stacktrace? That snippet does not contain enough info. >> >> Thanks, >> Hari >> >> >> On Fri, Oct 31, 2014 at 4:16 PM, terreyshih <[email protected] >> <mailto:[email protected]>> wrote: >> >> Hi, >> >> >> Can somebody help me explain the exception ? >> >> An exception was thrown while closing send thread for session >> 0x248485dc55fe84b : Unable to read additional data from server sessionid >> 0x248485dc55fe84b, likely server has closed socket >> >> thx >>
