When we try to render C4D Arnold on the farm, it will render one frame and hang up on the next. It doesn’t report any errors and in the worker logs, it just sits there and hangs. It never reports a completion percentage. Our workaround is to set a task timeout so it’ll requeue and try again. Sometimes we’ll monitor the job as it’s running and manually requeue tasks that have gone a few minutes and haven’t reported a completion percentage yet, so we don’t waste time waiting for the timeout.
If you look at the logs, you can see how it’ll render and then hang, render and then hang. Keeps going back and forth from one task to the next, even on the same node.
We’ve only seen this particular hang up happen with Arnold, never with Redshift. Our C4DtoA version is v3.2.1. It doesn’t seem to be tied to a particular version either because it was happening with prior versions since Fall last year.
Has anyone seen this before? Any advice would be very much appreciated. Thanks.
Just checking in. Anyone have any ideas? It’s still happening on our farm and is really frustrating and affecting our deliveries. Appreciate any advice.
Hello!
We’d have to see some logs from the hanging tasks, which will likely require jumping onto the Worker that was hanging onto the job.
The logs will be on the machine in one of these locations:
Windows: C:\ProgramData\Thinkbox\Deadline10\logs
Linux: /var/log/Thinkbox/Deadline10
Mac OS X: /Library/Logs/Thinkbox/Deadline10
The forums don’t have a mechanism to ensure regular follow-ups, but the ticket system at awsthinkbox.zendesk.com/ does, so I’d like you to create a ticket there instead so you’re not left out to dry again.
2 Likes
Update. Submitted a ticket and still working with support on this issue. So far we don’t know what’s causing this, but we do have a workaround. The jobs need to be submitted with “Use Batch Plugin” unchecked. This forces the scene to be reloaded into memory on each task, and it looks like it fixes the issue. So, just in case anyone comes across this same problem, that solve is working for us for now. If we get a permanent solution, I’ll follow up again here. Thanks.
