AWS Thinkbox Discussion Forums

Job loop again and again....

Hi we are having problem with sim and maxscript jobs.

Amy clue would be great. I’ll attach log file later.

We are having an issue with Network Simulations that get sent to Deadline to process.
They just continuously write cache files and do not stop.

This was first discovered with FumeFX. Now we are seeing the same behavior with Xmeshes sent to the farm for Caching.

Here is a job on Deadline that is a network Xmesh simulation cache:

This job is 225 frames of cache. Once it gets to the end it will just continue to repeat and start again at frame 1. So it doesn’t stop when it reaches 225.

Please always provide the exact version of Deadline, 3ds Max, and the plugins (XMesh, FumeFX) you are running so we can better replicate.

Without a log, we cannot tell anything, but it sounds like the job is not finishing, so it is restarting from the beginning.

  • Does the log end with the lines

0: INFO: XMesh Saving Frame : 225.0 0: INFO: Finished XMesh Saving Job in XX.XXX sec. 0: Done executing plugin command of type 'Render Task'

  • Do you get the Error counter going up on the job and task?

I just submitted a test using Max 2017, XMesh 1.4 and Deadline 9.0.1.1, and the job finished as it should.

I cannot wait to see the log of a machine that constantly requeues the job, as it might show us where it is failing.
The fact it is also failing with FumeFX simulation jobs (which script did you use to submit that?) shows that it is not an XMesh-specific issue, but most probably 3ds Max failing for some reason and the Deadline Slave catching that and retrying.

I attached a log from xmesh cache job.

It ended with…

2017-04-10 16:27:36: 0: INFO: Finished XMesh Saving Job in 7807.71 sec.
2017-04-10 16:27:36: 0: Done executing plugin command of type ‘Render Task’

But… i’m seeing…

=======================================================
Reason

Rendering task was requeued because the Slave was manually shut down.

No one touched that job at the moment.
Xmesh.txt (141 KB)

For FumeFX submission, we have own script.
Basically, we submit as a frame render job with BackBurner button On in FumeFX.

Thank you for the log! I have asked our support team to take a look at it.

To me, it appears that the Slave is shutting down for some reason.
Can you white-list the job to only one Slave, or even your own workstation, and watch what the Slave is doing while processing the job?Does the Slave actually restart?

Checked the source code… It looks like that message is a generic catch-all for when a job is re-queued.

I’m wondering if something may be corrupt in the binaries folder. It wouldn’t hurt to upgrade to SP17 either. Would you be willing to give that a shot? If it still hits, it would be great to get the full Slave log over in the support system so I can take a look at what’s happening. Also, if you guys are running Pulse, that would be good too as it may be something strange with house cleaning.

What’s SP17?
How can I get the full Slave log?

Here’s some info on grabbing the Slave logs: (paths are shown further up the doc)
docs.thinkboxsoftware.com/produc … html#slave

Service Pack 17 is the last release of 8.0, so you can do a minor upgrade to get to that version. Info on minor upgrades here:
docs.thinkboxsoftware.com/produc … downgrades

To get the installers, use the download link Sales provided along with your license or e-mail them back asking for one.

Privacy | Site terms | Cookie preferences