Duke-UNC Brain Imaging and Analysis Center
BIAC Forums | Profile | Register | Active Topics | Members | Search | FAQ
Username:
Password:
Save Password   Forgot your Password?
 All Forums
 Support Forums
 Cluster Support
 cluster logs?
 New Topic  Reply to Topic
 Printer Friendly
Author Previous Topic Topic Next Topic  

clithero
Junior Member

37 Posts

Posted - Nov 15 2007 :  10:29:11 AM  Show Profile  Reply with Quote
Some of my scripts are not working and the bugs are eluding me. Is there some reason not all of my scripts are outputting log files?
Thanks,
John

deshmukh
BIAC Alum

40 Posts

Posted - Nov 15 2007 :  10:49:45 AM  Show Profile  Reply with Quote
Are you having problems with multiple jobs using the same script i.e. some of the jobs using the script run ok but others fail or are these different scripts ?

You might want to do a basic syntax check on scripts by running
sh -n script

If the script passes that test without errors, then check to see if all the data is in the expected location. This tends to be the most common issue.

The job level log files are generated on job completion. If the job dies due to command errors (usually generated by incorrect paths) then these files are never generated. You can however debug scripts by adding your own debug messages in the "USER SCRIPT" section to a specific file in your Experiment folder. For eg. here is the most common debug message,
echo "GOT HERE" >> $EXPERIMENT/Analysis/Logs/debug.txt

You can always add your own tags to messages so you can identify which job generated a particular message eg:
echo " $EXAM_NUMBER : Running feat" >> $EXPERIMENT/Analysis/Logs/debug.txt


















Go to Top of Page

clithero
Junior Member

37 Posts

Posted - Nov 15 2007 :  11:37:00 AM  Show Profile  Reply with Quote
Thanks, I will use that check.
However, it appears that the cluster may not be accepting new jobs. Multiple people have tried running scipts that worked fine before, but now are not running. Node 7 in particular. The jobs seem to run for about 5 minutes and then quit without error. The scripts themselves are (as far as I can tell) bug free.
Go to Top of Page

deshmukh
BIAC Alum

40 Posts

Posted - Nov 15 2007 :  11:55:58 AM  Show Profile  Reply with Quote
I was able to submit a job to the cluster without problems so the cluster is accepting jobs. The node7 problem was related to experiment access and is fixed now.
Go to Top of Page

wang
BIAC Alum

USA
57 Posts

Posted - Nov 28 2007 :  5:38:41 PM  Show Profile  Reply with Quote

Does anyone have problem with running scripts in the Cluster now? A script that worked previously does not work now. Here is the error message for the Log file:

Tcl_InitNotifier: unable to start notifier thread

Have we already upgraged fsl to 4.0.2? Thanks.

lihong
Go to Top of Page

Carpenter
New Member

16 Posts

Posted - Nov 28 2007 :  5:49:31 PM  Show Profile  Reply with Quote
I just got the same error...
Go to Top of Page

josh.bizzell
BIAC Staff

USA
118 Posts

Posted - Nov 29 2007 :  09:55:28 AM  Show Profile  Reply with Quote
We did not update FSL to 4.0.2 on the cluster just yet; we're waiting to make sure it is completely bug-free before doing that.

This error seems to have been caused by a change in the configuration we made to the grid engine late yesterday. We've reverted back to the original setting and now it seems to work.

-Josh
Go to Top of Page

tankersley
BIAC Alum

USA
143 Posts

Posted - Jan 04 2008 :  3:50:18 PM  Show Profile  Reply with Quote
I am trying to get my first cluster script running. I have run sh -n script.sh on it and do not get any error messages.
I've also inserted the echo "GOT HERE" command as the first line in USER SCRIPT, but the job never gets to this portion of the script.
Also, the job keeps exiting with exit code "1", which means(?) that I'm misusing a built-in shell command.

Any help appreciated.

Thanks,
Dharol
Go to Top of Page

dvsmith
Advanced Member

USA
218 Posts

Posted - Jan 04 2008 :  4:38:03 PM  Show Profile  Visit dvsmith's Homepage  Reply with Quote
I was having exit code "1" pop up when node6 wasn't working earlier today, but I thought that was fixed. If your jobs are going to node6 (should be able tell from the email) and you're not getting a log file, try a different node.

qsub -l hostname=node7 -v EXPERIMENT=whateverexp whateverscript.sh
Go to Top of Page

tankersley
BIAC Alum

USA
143 Posts

Posted - Jan 07 2008 :  12:29:55 PM  Show Profile  Reply with Quote
FYI, the problem appears to have been that the flag -v is required to run any script.

dharol
Go to Top of Page

syam.gadde
BIAC Staff

USA
421 Posts

Posted - Jan 07 2008 :  12:44:09 PM  Show Profile  Reply with Quote
Just to clarify Dharol's observation. It's not only that -v is necessary for scripts that access data on BIAC file servers -- it's that if you are specifying a variable to qsub (like EXPERIMENT=Dummy.01), you do that on the command line by using -v, so
in:

qsub -v EXPERIMENT=Dummy.01 ...

-v must be right before EXPERIMENT=Dummy.01
Go to Top of Page
  Previous Topic Topic Next Topic  
 New Topic  Reply to Topic
 Printer Friendly
Jump To:
BIAC Forums © 2000-2010 Brain Imaging and Analysis Center Go To Top Of Page
This page was generated in 0.54 seconds. Snitz Forums 2000