SCSM 2012 Cube Jobs Start but Never Finish

Occasionally my SCSM 2012 cubes, either started manually or kicked off by scheduled overnight jobs, would never stop. It’s like the handshake between the console and the SSAS server fails.  Keep in mind that running cubes from the SCSM 2012 console (either manually or automatically) take 2-3 times longer than running them from the SSAS, but shouldn’t take hours to complete.

To help me establish a run time baseline, I manually ran all of my cubes with a PROCESS FULL criteria and recorded the times. None of my manual processing took over 5-minutes, so any job running over an hour from my console was probably a dead job. This was my approach to fixing this problem. I can’t say this enough, evaluate it in your test environment:

  1. Stop the job at the console.
  2. Check the StatusID (as described by Danny Chen) of the cube and modules. They should all be stopped.
  3. Reset the StatusID of any stalled job and module to 3 (Not Started)
  4. Manually Process Full the failed cube at the SSAS to ensure you have no failed DIMS. Any error message shows the failed DIM.
  5. Process failed DIMS by a) Unprocess b) Process Full.
  6. Try processing the Cube again. If successful, un-process the Cube. So, a) Process Full b) Good Run c) Unprocess
  7. Reset the watermarks for only the failed cube to 0.
  8. Manually run the cube job from the Data Warehouse via PowerShell.
  9. If it still keeps running, start at step 1 and try again, but skip step 6. Sometimes leaving the cubes Processed before a manual start fixed the jobs.

Step 4 is where I would find the root cause of my occasional cube failures. SCSM 2012 isn’t very adept at communicating bad DIMS back to the console and this appears to simply stall the job in a running state. The DIM fix suggested by Thomas Ellermann’s blog post is an “all or nothing” approach that will usually correct cube processing problems, but wouldn’t fix my stuck jobs. For my issue, I had two bad cubes and this seemed like overkill to me to reprocess all of the DIMS, so processed cubes manually at the SSAS, although Microsoft doesn’t recommend it since this will place the Data Warehouse watermarks out of synchronization. Should you decide to manually process, it’s important that you update your watermarks as described above.

After un-processing the bad DIMS, I manually processed the DIMS and then manually processed the cubes via SSAS. I verified both of the cube jobs by running them via PowerShell with no issues. The next morning I found that all cube jobs ran successfully and my cubes have been stable ever since.

DS

Advertisements

About Doug Sigmon

IT Helpdesk manager in southern California. Love technology, gadgets, and golf.
This entry was posted in SCDW, SCSM 2012 and tagged , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s