Sorry for the thread necromancy, but did you find a solution to the 'refresh' problem, i.e. having a coordinator automatically pick up a changed workflow (and properties) for the next materialization? This seems such a fundamental issue that I'm hesitant to roll out my own wrapper that manages this.
Thanks, Abhinav On 1 April 2014 at 02:49, <[email protected]> wrote: > Hi, > > I'm having trouble understanding "rerun" (and refresh) in Oozie when it > comes to coordinator jobs. > > Here's what I'd like to do - please let me know what the right way to do > this in Oozie. > > Suppose at time A I submit a coordinator that will periodically runs some > workflow. At time B, I decide to make some changes to either the workflow, > the coordinator or the underlying code, so I redeploy the code, > workflow.xml, > job.properties, and coordinator.xml into HDFS at the same location I had it > before. > > What should I do with the coordinator that is running? Should I kill it > and just do a -run to create a new one? I tried suspending it and resuming > it but it doesn't seem to reread the coordinator.xml file (where I changed > the start time for example). > > I tried killing the coordinator and then doing a -rerun (or -refresh > -rerun), but I got the error message: E1018: Coord Job Rerun Error: > coordinator job is killed or failed so all actions are not eligible to > rerun! > > Ideally, I like to "refresh" the coordinator and tell it that things have > changed and that it should reread its coordinator.xml and job.properties > and perhaps redo its schedule for the workflows, etc. > > What's the right way to do this? > > Thanks, > Michael >
