AWS Thinkbox Discussion Forums

Nuke jobs hanging indefinitely

Deadline 6.2 Beta 5
Nuke 7.0v8 x64
Windows 7 x64

We have a bunch of clones that we’ve moved offsite, they render Max scenes fine, but when it comes to Nuke they seem to render through 100% and then hang, completely finished with the job they just idle and stop reporting their status and anything else. In the monitor I believe it just shows them still working on the job and it apparently bypasses any timeout at that point since the task will sit there completed on the machine for days.

There is no hardware or software difference as these machines are the same clones we use locally on-site. The only difference is the network, these are at a remote site that is approx 1ms further away than our local LAN latency (usually somewhere below .5ms). The route is a little larger of course (goes through 6 devices), but our 3dsMax 2014 scenes are rendering fine.

Any idea where to start troubleshooting this?

Thanks guys,

Joe

Hey Joe,

The first thing to try is to resubmit the Nuke job with the “Use Batch Mode” option disabled. When disabled, Deadline will just perform a regular Nuke command line render, instead of keeping Nuke active between tasks. If you still run into the same problem, grab us the log from the slave so we can see what it’s doing when it gets stuck.

Thanks!
Ryan

We have the same problem with non-nuke-batch mode renders. Sadly going to batch mode is non-trivial in our case - although on the long term todo list.

I believe this is due to nuke not flushing its std out properly, and getting into a deadlock.

FYI.
I have a Foundry support ticket since last December, asking for STDout in Nuke on Windows to be improved, as it can slightly knock out our progress reporting and is visible in the slave logs.
2…3…4…5…6…7…8…9 text can sometimes appear out of sync with the rest of the STDout due to Nuke’s irregular STDout ‘flushing’, particularly visible on the Windows OS platform.

Privacy | Site terms | Cookie preferences