Hey all,
Last night we upgraded from oozie ~2 (cdh3) to oozie 3.2.0 (cdh4), and we're seeing some really concerning behavior that we can't explain. Whenever a workflow is submitted, it will run through each action as normal except that when each action begins it will spend 15 - 45 minutes in the 'PREP' state. Clearly this really slows things down. This seems to be happening across the board for every workflow. I can't see anything particular in the oozie server logs that might help, and our settings are almost identical to the working oozie2 server we had working only two days ago. The load we're putting on the server is minimal compared to our usual production workload. I'm hoping that some of the experts in the community might be able to help me debug this, or hopefully have seen something similar before? In case it helps, we're deployed on Amazon AWS, and we're using Amazon RDS for the oozie database. Thanks for any help you might be able to offer. -- Matthew Rathbone Foursquare | Software Engineer | Server Engineering Team [email protected] | @rathboma <http://twitter.com/rathboma> | 4sq<http://foursquare.com/rathboma>
