At my client location they have turned on capacity scheduler , oozie job was initiated, first part it sqoops the data then a pig job is run. Sqoop job was successful, then pig job is initiated but without 0% progress, any pointers ?
PErhaps Memory issues? I.e. only the AM is started but no room for tasks? Or in other words how much RAM do you have in your queue and how much is
oozie launcher am
oozie launcher mapper
pig am
together using. If there is no space anymore for any other mapper it will wait forever. But you can see this in Resourcemanager. ( if the pig task has mappers scheduled ). Then its a different problem.
I checked the memory, it doesn't seem to be a problem. same job runs fine with Fair Scheduler. Is this could be an issue with HA ? I went thru the logs, state is updated from time to time launched to running, then it keeps on looping from that point.
check the individual nodemanager logs where tasks fail. I would go through all Capacity Scheduler properties again and make sure FairScheduler policies do not conflict with it. @kjilla
