Hey Guys,
Due to a log4j version conflict of elasticsearch 5  and storm1.* log4j
version, I manually replaced the log4j version in storm lib folder to 2.7.

 log4j-api-2.7.jar     log4j-slf4j-impl-2.7.jar      slf4j-api-1.7.7.jar
log4j-core-2.7.jar    log4j-over-slf4j-1.7.22.jar

However, after I replaced the jar, when running in local mode, it seems to
working fine, but once deployed, the topology is shown in the UI, but does
not seem to be live. I checked nimbus log, and there are lots of executor
not alive message:

2016-12-31 01:48:03.950 o.a.s.d.nimbus [INFO] Executor
staging_event_topology-3-1483148476:[2 2] not alive
2016-12-31 01:48:03.950 o.a.s.d.nimbus [INFO] Executor
staging_event_topology-3-1483148476:[3 3] not alive
2016-12-31 01:48:03.950 o.a.s.d.nimbus [INFO] Executor
staging_event_topology-3-1483148476:[1 1] not alive
2016-12-31 01:48:03.950 o.a.s.d.nimbus [INFO] Executor
staging_event_topology-3-1483148476:[6 6] not alive
2016-12-31 01:48:03.950 o.a.s.d.nimbus [INFO] Executor
staging_event_topology-3-1483148476:[5 5] not alive
2016-12-31 01:48:03.951 o.a.s.d.nimbus [INFO] Executor
staging_event_topology-3-1483148476:[4 4] not alive
2016-12-31 01:48:03.957 o.a.s.s.EvenScheduler [INFO] Available slots:
(["0c9066aa-93ca-4161-8069-3ae41269369e" 6702]
["0c9066aa-93ca-4161-8069-3ae41269369e" 6703])

Also no log is generated for the topology at all.

Once I restored the log4j back(2.1), it starts to working again.

Anyone has seen this before?

Thanks!
Chen

Reply via email to