Search code examples
hadoophbasedistributed-computingoozie

Can not kill oozie job by -kill command


I tried to kill a oozie coordinator job like this:

$ oozie job -oozie http://10.0.3.2:11000/oozie -kill 0003288-130913181709024-oozie-oozi-C

There is no any error appears.

But after I check the oozie UI, the job still exists.

I skilled severa jobs , then deployed new code update on oozie , the start new job. Since the old job can not be killed, there are many jobs with same project'RUNNING' on Oozie now.

I can kill the jobs before , but can not kill it now. So, how to force kill the RUNNING job ? Do you know what may made this happen?

Thanks very much.


Solution

  • I encounter this error sometimes on my testing environment, where Oozie uses Derby as database. The solution is to clean Oozie state by removing the database:

    sudo /etc/init.d/oozie stop
    sudo rm -rf /var/lib/oozie/oozie-db/
    sudo /etc/init.d/oozie start
    

    Of course, this solution may not be appropriate for production system (althrough I've never seen this error in production).