It would be great if Deadline didn’t throw away logs for tasks that time out, as it’s helpful to see what was going on when they were killed.
Thanks.
It shouldn’t be. Is this happening with Deadline 6 or 7? Can you send us a timeout error report so that we can see what’s in it?
Thanks!
Ryan
This is using 6.2. Here’s what I get in the report log:
[code]=======================================================
Error
The Slave did not complete the task before the Regular Task Timeout limit of 00d 06h 00m 00s. The Task Timeout settings can be changed for this job by right-clicking on it in the Monitor and selecting “Modify Properties…”.
=======================================================
Type
JobTimeoutException
=======================================================
Stack Trace
=======================================================
Full Log
0: Slave timed out – canceling current task…
=======================================================
Details
Date: 09/11/2014 05:13:49
Frames: 1050
Elapsed Time: 00:06:00:03
Job Submit Date: 09/10/2014 17:12:05
Job User: alexc
Average RAM Usage: 3986246892 (12%)
Peak RAM Usage: 4133658624 (13%)
Average CPU Usage: 1%
Peak CPU Usage: 22%
=======================================================
Slave Information
Slave Name: rd-083
Version: v6.2.0.32 R (2563d5bc8)
Operating System: Fedora release 19 (Schrödinger’s Cat)
Machine User: qmaster
IP Address: 100.100.103.83
MAC Address: FF:FF:FF:FF:FF:FF
CPU Architecture: x86_64
CPUs: 24
CPU Usage: 3%
Memory Usage: 3.7 GB / 31.4 GB (11%)
Free Disk Space: 69.316 GB
Video Card: Matrox Electronics Systems Ltd. MGA G200eW WPCM450 (rev 0a)[/code]
Same story for two different tasks on this job, executed on two different machines.
Thanks! That’s definitely a bug. We’ve logged it and will look into it.
Cheers,
Ryan
Does this happen whenever a timeout occurs?
No, not always. I’m still sort of wondering if it’s related to the other issue of slaves having issues with really large amounts of task output (thread here: viewtopic.php?f=86&t=12347).
Also, I’ve seen some task logs in the repo where the first 70-80% of the log was just not there. These tasks didn’t time out, but most of their actual log was missing.