Job_2017-10-20_09-21-40_59ea229440ca65581c3b9cef.txt (47.7 KB)I’m having a problem with an Arnold scene running on our farm. I’m running multiple slave instances on one of our blades. If I remove all the instances and just have a single slave on the blade the frames render just fine. But if I add any more instances then every task will fail. I’ve attached the log file for a failed render.
Hi Charles. So I went ahead and submitted another job with the Maya Batch turned off and I also upped the verbosity. At first I got a good chunk of frames that rendered just fine but then the errors started coming again. I’m basically rendering on 3 slave instances from one blade. Some successful frames still get through, but it seems like they fail more often than they succeed. I also almost seems like it fails when two or more tasks are trying to write out to disk at the same time. I’m attaching two job reports that failed at around the same time and one job that succeeded.
EDIT: Actually, it looks like the success to fail ratio is closer to even.
EDIT 2: After about 30 frames it seems to have stabilized. No errors since the first batch. Weird.
It’s good stuff, but if it’s happening across multiple Slaves at the same time that could be hard to reproduce. I did find this guy, but I assume it’d throw that error every time: forums.autodesk.com/t5/maya-sha … -p/7013692
So, I’m suspecious of OpenEXR’s error here. This is a shot in the dark, but if the file server is overloaded could you try enabling “local rendering”. It’s under the “Maya Render Job” rollout.
Hey Edwin, sorry for the delayed response but I was away on PTO late last week.
So yeah I came around the the local rendering solution myself. it seems to alleviate the issue where Maya has trouble writing out the render file to disk. This has been a weird error to debug since other test scenes using Arnold seem to render without issue. I’m going to continue to dig into the bad scene in hopes to identify the problem. I’ll check out the links you provided.