AWS Thinkbox Discussion Forums

Machine Error Limit

Hi,

We are running Deadline 10.1 with Cinema4d (S24) and Redshift (3.0.67)

Sometimes a particular machine might begin to error due to gpu memory, but it then ‘runs away’ through the rest of the frames, erroring them all out and thus failing the job.

Is it possible to exclude the erronous machine based on a limit. So if that particular machine fails to render a frame more than 5 times due to catastrophic errors, it is kicked off the job for good so the other machines can complete their tasks?

Any and all info much appreciated!

Kind Regards,

C

yes, job failure detection will do this
https://docs.thinkboxsoftware.com/products/deadline/10.1/1_User%20Manual/manual/failure-detection.html
you can also set a task timeout minimum, so it’ll fail too quick renders also
https://docs.thinkboxsoftware.com/products/deadline/10.1/1_User%20Manual/manual/job-controlling.html#timeouts

2 Likes
Privacy | Site terms | Cookie preferences