Mailing List Archives
Authenticated access
|
|
|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Condor-users] Simple DAGman problem
- Date: Tue, 28 Mar 2006 07:27:24 +1100
- From: "Mark Van De Vyver" <mvyver@xxxxxxxxx>
- Subject: Re: [Condor-users] Simple DAGman problem
Kent,
thanks for the corrections.
I've made changes along the lines you suggest. Unfortunately I now
get the error:
Checking all your submit files for a consistent log file name.
This might take a while...
ERROR: submit files use different log files (detected in: upload.)
Both submit files have the same log file defined (the log file does
not already exist on in the log files directory). I include them and
the DAGman file below:
# ************************** GenData_R.dag **************************
JOB Inspect4R InspectForR.sub
JOB RunRJob GenData_R.sub # uses RPresent env variable, set by
InspectForR post script, to select file to upload.
SCRIPT POST Inspect4R Run_SetRPresent $RETURN
PARENT Inspect4R CHILD RunRJob
QUEUE
# ************************** InspectForR.sub **************************
Universe = vanilla
Requirements = (MEMORY > 64) && (Arch == "INTEL" )&& ( OpSys ==
"WINNT50" || OpSys == "WINNT51" )
notification=Error
notify_user=mvdv@xxxxxxxxxxx
Output = R\output\condor_out_$(Cluster)_$(Process).txt
Error = R\error\condor_error_$(Cluster)_$(Process).txt
Log = R\log\condor_log.txt
should_transfer_files = YES
transfer_files=ALWAYS
WhenToTransferOutput = ON_EXIT
transfer_input_files = R\input\Run_InspectForR.cmd, R\input\Run_SetRPresent.cmd
on_exit_remove = (ExitBySignal == False) && ((ExitCode == 0)||(ExitCode == 1))
Executable = R\input\Run_InspectForR.cmd
Queue
# ************************** GenData_R.sub **************************
Universe = vanilla
Requirements = (MEMORY > 64) && (Arch == "INTEL" )&& ( OpSys ==
"WINNT50" || OpSys == "WINNT51" )
Getenv = False
notification=Error
notify_user=email@address
Arguments = $(CLUSTER) $(PROCESS) 10 GenerateSimulatedSeries_MySQL.R
Output = R\output\condor_out_$(Cluster)_$(Process).txt
Error = R\error\condor_error_$(Cluster)_$(Process).txt
Log = R\log\condor_log.txt
should_transfer_files = YES
transfer_files = always
transfer_input_files = R\input\R-2.2.1_$ENV(RPresent).exe,
R\input\GenerateSimulatedSeries_MySQL.R, \
R\input\Run_SetRPresent.cmd
when_to_transfer_output = ON_EXIT
on_exit_remove = (ExitBySignal == False) && (ExitCode == 0)
Executable = R\input\Run_GeneratesimulatedSeries_MySQL.cmd
Queue
I will appreciate any suggestions you may have.
Regards
Mark
On 3/28/06, R. Kent Wenger <wenger@xxxxxxxxxxx> wrote:
> Mark,
>
> > Thanks to the developers for all their efforts.
> > I'm running condor version:
> > $CondorVersion: 6.6.10 Jun 22 2005 $
> > $CondorPlatform: INTEL-WINNT50 $
> >
> > I seem to be doing something simple wrong: I have two condor job
> > files that run fine when run separately.
> > However, if I call them using the following DAGman file I get an
> > message from Condor that doesn't indicate anything is wrong, but the
> > jobs don't get submitted, in fact using condor_submit_dag -no_submit
> > doesn't result in the intermediate condor.sub file that the
> > documentation mentions.. The Condor message is below the Dagman file:
> >
> > JOB Inspect4R InspectForR.sub
> > JOB RunRJob GenData_R.sub
> > VARS RunRJob RPresent = $RETURN
> > PARENT Inspect4R CHILD RunRJob
> > Queue
> >
> > Checking all your submit files for a consistent log file name.
> > This might take a while...
>
> I'll be the problem is the fact that you don't have double quotes around
> your value in the VARS definition, e.g.,
>
> VARS RunRJob RPresent = "hello, world"
>
> Also, $RETURN is only defined for POST scripts, not jobs, because it's
> the return value of the job.
>
> I guess condor_submit_dag should generate some error messages... I'll
> check whether it does in 6.7, at least.
>
> > I should note that there is no error, log or output files defined in
> > either of the two job files used in the DAGMan file.
>
> That's another problem -- the node jobs *must* define log files -- if they
> don't, DAGMan has no way to monitor their progress, because it does so by
> reading events from the log files.
>
> Kent Wenger
> Condor Team
>
--
Mark Van De Vyver, PhD
--------------------------------------------------
My research is available from my SSRN Author page:
http://ssrn.com/author=36577
--------------------------------------------------
Finance Discipline
School of Business
The University of Sydney
Sydney NSW 2006
Australia
Telephone: +61 2 9351-6452
Fax: +61 2 9351-6461