getting ready to start to render a huge amount of jobs via Deadline with Maya 2016.5 and Mental Ray 2016.5 using all Mila Shaders. The Render came out fine, but did not render the first 11 frames which looks as if it was a buffer or file latency issue (but still said Rendered 100%) and then another 45 pr so images in the 1890 frames do not know which ones “rendered 100%”, but did not output an image. What can I do to fix this? Is there a way to show which frames were actually not output-ted? I am in gearing up testing to make sure rock solid phase and this is a little confusing…Thanks
Assuming Deadline knows the output path, you should be able to right click the job in the Monitor, go to Job Output, then scan for missing output. This will tell you which output is missing, and give you the chance to requeue the related tasks. If the slave doesn’t know that the file didn’t copy to the output location, Deadline won’t be able to report it as bad. Can you share a job report from one or two of these instance where the task was marked completed, but the output didn’t get written? Thanks.
Ahh if there is no Job Output option in the menu after right clicking the job, Deadline doesn’t know where the output is, so it cannot check it. If you provide the output path as part of your submission, the Scan option will be available.
The path was set and the same computers rendered out other images just fine, its a strange phenomenon. Even with the legacy layers turned on, it still did the same thing, I had to baby sit and recheck as it was rendering and resubmit tasks as they came through.
Can you share job reports from these tasks that are failing so we can try to see what is being recorded as the cause, and thus allow us to code in an error detection, and thus failing the render.
Unfortunately by your description it sounds like Maya is exiting with a code 0, even though the render was a failure, which, when coupled with the fact we aren’t catching an error based on our code, is akin to Maya telling Deadline ‘Hey, all’s good here!’ and Deadline now moving on to the next task, because it detected no errors, and the application completed with an exit code, or in the analogy a message, denoting success.
Hopefully with some job reports we can figure out what is causing the failure, or at least something in the log we can use to detect the failure, and at least the failed attempts will be marked as the failures they factually are.
This is the same issue in your other forum post regarding maya 2016.5:
In Maya 2016.5, Autodesk introduced a new render layer system which is incompatible with their old one. The system is loaded on start up via the preferences on a per machine basis. So the machines that are succeeding have it set to load to the Legacy Render Layer system, whereas the slave from this log hasn’t had that preference changed (I believe it defaults to the new “Render Setup”).
From the log file, it seems that you’re on version 8.0.0.69. In version 8.0.3, we added support to handle this by switching slaves to the proper version of render layers for the file they’re rendering (we gather this info during submission). The easiest solution, if possible, would be to upgrade to the latest version of Deadline, since we handle this for you.
However, if you’re unable to upgrade right now but want this problem resolved you can have a few options available to you:
If you have files that use the new system and files that use the old system, then you could create groups for each render layer system and assign them on submission.
open maya up on the slaves giving you trouble and change their preference to the Legacy system
blacklist the slaves giving you trouble
use 2016 if you don’t specifically need 2016.5 features
Thank you, but why do they render out 90% of the sequence? I dont understand why it would render x amount of frames and not the other 10%. We did change it to the legacy render settings, it still gave the same issues, but that was prior to me upgrading to the new build, so I have to test it today. Thank you for your diligence.