[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: [Condor-users] performance issue....
- Date: Wed, 20 Oct 2004 07:25:21 -0700
- From: "bruce" <bedouglas@xxxxxxxxxxxxx>
- Subject: RE: [Condor-users] performance issue....
i looked/reviewed the example submit files in the docs, although i haven't
actually set up the situation that i'm describing. what i am trying to get a
better understaning of, is how/what kinds of information willl i have in the
log files for a given app, if i run a number of different apps at the same
time, via a single submit file.
in the example that you pointed me to, it uses a single log file. so in that
log file, would there be unique identifiers within the file, that i could
trigger off of to caputre information for the app/process being run.
ie. if i had perl scripts foo.pl, and cat.pl, and ran them from the same
submit. would the log file have the name of each perl app that was run, so
that i could extract information on a app/process basis..
or would the log file have some unique identifier, but not neccessarily the
[mailto:condor-users-bounces@xxxxxxxxxxx]On Behalf Of matthew hope
Sent: Wednesday, October 20, 2004 1:03 AM
To: Condor-Users Mail List
Subject: Re: [Condor-users] performance issue....
On Tue, 19 Oct 2004 11:18:50 -0700, bruce <bedouglas@xxxxxxxxxxxxx> wrote:
> ok.. one more question that i forgot! if i run multiple apps from a single
> submit as my example, will the log file show information for each app, or
> for each submit job??
> if the log file essentially only gives information at the job level
> of for each app being run within the job, then i could take advantadge and
> just have lots of submit jobs, one for each app i have to run...
you do not have to be constrained in that way even if you use one
submit file for each 'class' of jobs.
if you extend the examples at
to include the $(Cluster) macro then you can set
log = MYLog.$(Cluster).$(Process).log
or somesuch, then everything gets a different log (until you change
the schedd so that the clusterids are no longer unique but that should
be a reasonable amount of time and you can always force through noop
jobs till the number is monotonic.
Condor-users mailing list