AWS Thinkbox Discussion Forums

Nuke submission script not starting

Hi guys.

I keep on having problem with the submission scripts in our networked environment. Latest one is with Nuke :laughing:
I’ve tried installing it manually, adding it to the plugins dir of the Nuke installation (not /home/profile/.nuke) and adding the proper lines in menu.py which is in the same folder. This creates the Thinkbox dropdown menu in Nuke, but clicking on the button doesn’t do anything. It’s just like the script does not start at all.

Weird thing is that installing it manually via the installer doesn’t work at all too, as the installer reports that its unable to find the repo, even after I’m pointing it to the Deadline bin folder. Bizarre!

Hey excessive,

What Deadline version are you on? Is there a particular reason you did not install it to the .nuke folder?

Can you post the full error you’re receiving? You 'll get the most info by opening Nuke’s Script Editor panel and then trying to launch the submission script.

As for the installer issue, can you confirm that the environment variable DEADLINE_PATH is set?

Cheers

Hey!

I wrote up a little response to this. Here’s my canned response. :smiley:

Nuke’s integrated submitter is written in Python and has a console for error reporting. Surpringly it’s named “Script Editor” and the error console will not contain error messages. To view the console, right click on a tab in the interface and choose the “Script Editor”. Messages from our submitter should be shown there.

[attachment=0]Nuke Script Editor.png[/attachment]

Can you check for error messages there? They should be helpful.

Update: Morgan beat me to it!

Hey guys. Indeed, that was quite useful!

This is what is reported

/Volumes/studio/software/nuke - thats obviously the location of our networked Nuke install.

path = GetRepositoryPath(“submission/Nuke/Main”) - this makes me think that it is looking for the Main folder of the Deadline repository at wrong place. The repo path is /Volumes/studio/deadline/repo. I’m wondering If I change it to the whole path it will work?

Hmm, to me it looks like it’s failing to find deadlinecommand.exe (it’s failing to find the executable used by the subprocess).

Can you check what the path for DEADLINE_PATH is? You can find that out on OSX by checking the value in the file: /Users/Shared/Thinkbox/DEADLINE_PATH
I wonder if that path even exists for you. I could see if we’re unable to create it then it forgets where deadline bin is…

in which case, we’re gonna have to figure out a better solution to handle environments in OSX…
Just learnt a bit more about how OSX environment variables don’t really follow Windows or Linux. ie. how it changes between application types (gui, shell, daemon, non-interactive shell, etc.) and the nightmares associated with that. So the file makes a lot more sense. Let me know the status of the above file and we’ll go from there.

Cheers

Hi Mepp.

Managed to sort this out with simplest possible solution - by manually adding variables. So far the way I passed them to Deadline was through startup script which exports them and afterwards starts Deadline. Adding them manually to the system sorted this out.

Having the submission script work resulted in finding another issue. The job submitted through the script for some reason use only one Slave of the farm which is quite annoying. Submitting the same job with same settings from the monitor uses all of the available slaves. I hope that you can help me on this one as well.

Depending on your Deadline version (you’re good if you’re 10.0.10.4 or later), you can use a tool we built into the Deadline Monitor to help figure out why your job isn’t being picked up. Check out this blog post if your version is 10.0.10.4 or later. Otherwise…

Could you post the the submission params of both jobs? (In the Deadline Monitor, Right-click a job, select “Modify Job Properties”. The “Submission Params” Tab has all the info associated with job)

Values that would affect this are…

  • Machine List(Whitelist/Blacklist)
  • Machine Limit (if it’s not 0, it will limit the number of machines allowed to work on the job)
  • Pools
  • Groups
  • Limits (ie. license limits)

So there should be something that’s slightly different between the two of them, the submission parameters should show us what that is.

Cheers

Hi there.

Job Info parameters for the job submitted via the Nuke submitter

And the one submitted manually through the monitor

Hey,

Ahhhh, so it looks like you’re setting the Chunk Size (Frames Per Task) so high, that you only have 1 task to render.

ChunkSize=1000000 Frames=1001-1061 This means that the there’s only 1 slave that can render on that job, so you’ll want to specify “Frames Per Task” such that you have multiple tasks that can be worked on (you have ~60 frames you want to work on for the job, so make sure your frames per task is lower than that value).

You can change the currently submitted job by right-clicking the job, selecting “Modify Frame Range”, and setting “Frames Per Task” lower. To change it for any future jobs from Nuke, change the “Frames Per Task” value in the integrated submitter to something lower.

Cheers

Privacy | Site terms | Cookie preferences