Re: [Condor-users] Getting Matlab to run with an M-file on Condor

["Smith, Ian" <I.C.Smith@xxxxxxxxxxxxxxx>]

We've ran a lot of Matlab under Condor albeit using precompiled
executables and it can be *very* difficult to track these things down.

Once thing I noticed is that's it's exec'ing "C:\Program Files (x86)\MATLAB\R2007b\bin\matlab.exe"
- is that correct or should it be without the .exe as you'd type it on the command
line. Also does Condor parse that string OK with the space it ?

What is the default behaviour of this in the absence of an input file ? 
Does it exit straight away or pop up a window. In our Condor, under XP, 
popping up a window kills the job immediately but - here's the catch - *silently*. 
Also does the interpreted version of Matlab need to access any DLLs. 
If so can the job find these in it's path (this will usually be null by default) ? 
A DLL error will again pop up the window of death.

Our executables are generally wrapped in a .bat file and the only
way of getting to the bottom of some of the problems has been to
put a big sleep into it then login to the execute host and have a poke
around in condor execute dir. Try blowing away your environment so you
get the same kind of shell as the Condor job then try running the 
.bat file. Many times I've found the problem has been a missing DLL.

(More on the website: http://www.liv.ac.uk/e-science/condor/matlab/)

regards,

-ian.

------------------------------
Dr Ian C. Smith
e-Science Team,
University of Liverpool,
Computing Services Department


 

> -----Original Message-----
> From: condor-users-bounces@xxxxxxxxxxx 
> [mailto:condor-users-bounces@xxxxxxxxxxx] On Behalf Of Mills, 
> Ashley, VF-Group
> Sent: 31 August 2007 12:03
> To: Condor-Users Mail List
> Subject: [Condor-users] Getting Matlab to run with an M-file on Condor
> 
> I'm having a problem getting matlab to run which seems to be 
> very hard to diagnose. I'm using the following submit
> file:
> 
> ---- BEGIN SUBMIT FILE ----
> Executable = C:\Program Files (x86)\MATLAB\R2007b\bin\matlab.exe
> Universe = vanilla
> Log = test.log
> output = test.out
> error = test.err
> getenv = True;
> transfer_executable = False
> should_transfer_files = YES
> when_to_transfer_output = ON_EXIT
> initialdir = out
> transfer_input_files = t:\ashley\test\condor_test\mlab\j1\run.m
> Queue
> ---- END SUBMIT FILE ----
> 
> As you can see, I just want to run matlab with no arguments. 
> This is a test. I actually want to run it with arguments, but 
> when I run it with the argument "-r run" and run.m is one of 
> the files transferred, I get the same response as when no 
> arguments are ran: Matlab exits immediately and doesn't run 
> the m file, I know this because I made a really simple m file 
> that just tried to write to a file, and it didn't.
> 
> Here is the starter log:
> 
> ---- BEGIN STARTER.SLOT1 LOG ----
> 1 11:47:52 Starting a VANILLA universe job with ID: 108.0
> 8/31 11:47:52 Tracking process family by login "condor-reuse-slot1"
> 8/31 11:47:52 IWD: C:\condor/execute\dir_7052
> 8/31 11:47:52 Output file: C:\condor/execute\dir_7052\test.out
> 8/31 11:47:52 Error file: C:\condor/execute\dir_7052\test.err
> 8/31 11:47:52 Renice expr "10" evaluated to 10
> 8/31 11:47:52 About to exec C:\Program Files 
> (x86)\MATLAB\R2007b\bin\matlab.exe
> 8/31 11:47:52 Create_Process succeeded, pid=5992
> 8/31 11:47:52 Process exited, pid=5992, status=0
> 8/31 11:47:52 Got SIGQUIT.  Performing fast shutdown.
> 8/31 11:47:52 ShutdownFast all jobs.
> 8/31 11:47:52 ERROR: C:\condor/execute\dir_7052 still exists 
> after trying to add Full control to ACLs for PRIV_ROOT
> 8/31 11:47:52 **** condor_starter (condor_STARTER) EXITING 
> WITH STATUS 0
> ---- END STARTER.SLOT1 LOG ----
> 
> Here is the Shadow log:
> 
> ---- BEGIN SHADOW LOG ----
> 8/31 11:47:52 ******************************************************
> 8/31 11:47:52 ** condor_shadow (CONDOR_SHADOW) STARTING UP
> 8/31 11:47:52 ** C:\condor\bin\condor_shadow.exe
> 8/31 11:47:52 ** $CondorVersion: 6.9.3 Jun 13 2007 $
> 8/31 11:47:52 ** $CondorPlatform: INTEL-WINNT50 $
> 8/31 11:47:52 ** PID = 4336
> 8/31 11:47:52 ** Log last touched 8/31 11:47:27
> 8/31 11:47:52 ******************************************************
> 8/31 11:47:52 Using config source: C:\condor\condor_config
> 8/31 11:47:52 Using local config sources: 
> 8/31 11:47:52    C:\condor/condor_config.local
> 8/31 11:47:52 DaemonCore: Command Socket at <192.168.2.5:2683>
> 8/31 11:47:52 Initializing a VANILLA shadow for job 108.0
> 8/31 11:47:53 (108.0) (4336): Request to run on 
> <192.168.2.1:3439> was ACCEPTED
> 8/31 11:47:54 (108.0) (4336): ZKM: setting default map to (null)
> 8/31 11:47:54 (108.0) (4336): Job 108.0 terminated: exited 
> with status 0
> 8/31 11:47:54 (108.0) (4336): **** condor_shadow 
> (condor_SHADOW) EXITING WITH STATUS 100
> ---- END SHADOW LOG ----
> 
> So it all seems like there is no problem, the job is exiting 
> with status 0. But why is it exiting at all? If I run matlab 
> with no arguments it should never exit right?
> 
> Ashley Mills
> 
> _______________________________________________
> Condor-users mailing list
> To unsubscribe, send a message to 
> condor-users-request@xxxxxxxxxxx with a
> subject: Unsubscribe
> You can also unsubscribe by visiting
> https://lists.cs.wisc.edu/mailman/listinfo/condor-users
> 
> The archives can be found at: 
> https://lists.cs.wisc.edu/archive/condor-users/
>