AWS Thinkbox Discussion Forums

Slaves randomly hang at 100% complete

We have some jobs that are giving us some issues. About 10% of the slaves will render the frame completely, the task percentage will be 100%, but the slave will never fully release the task. We are seeing task render times around 10-15 minutes. The problem tasks will be at 100% complete and hang forever (we got impatient and killed it after an hour :slight_smile: ). It’s not confined to specific machines; meaning, a render node may render ten tasks fine, then get tripped up on a specific task. It doesn’t seem to be a specific section of the render; the problem is peppered throughout the job. We’re using MayaCmd.

Have you seen this problem before? If so, are there any pitfalls we may need to look out for? Are there certain conditions that exacerbate this problem?

Thanks!

It sounds like Maya is hanging at the end of the render. With the MayaCmd plugin, Deadline is just running the Render.exe process and waiting for it to complete, and my guess is that Render.exe just never exits in this case. If it’s limited to specific jobs, perhaps it’s something those jobs use (ie: a specific Maya plugin) that could be causing the problem.

I guess the first thing to do would be to post a log from the slave application when it gets stuck like this. Just go to the slave machine, select Help -> Explore Log Folder, and find the most recent slave log. We can look at it to see where Maya is getting stuck, and see if it’s logging any information that could explain what’s going on.

Cheers,

  • Ryan
Privacy | Site terms | Cookie preferences