On Thu, 22 Dec 2005, John Kennedy wrote:
> Hi,
>
> I am trying to modify the sge.pm jobmanager to fit to our cluster in
> munich but the gram_job_mgr_11128.log keeps getting filled with the following
>
> 12/22 14:59:22 JMI: in globus_gram_job_manager_poll()
> 12/22 14:59:22 JMI: local stdout filename =
> /home/cluster/uh351ba/.globus/.gass_cache/local/md5/92/f5245388ea5db7fcbe2768ca03ac9b/md5/5d/f67430397c6199ca8c23b152ebdf21/data.
> 12/22 14:59:22 JMI: local stderr filename = /dev/null.
> 12/22 14:59:22 JMI: poll: seeking:
> https://lcg-lrz-ce.lrz-muenchen.de:20011/11128/1135259481/
> 12/22 14:59:22 JMI: poll_fast: ******** Failed to find
> https://lcg-lrz-ce.lrz-muenchen.de/11128/1135259481/
> 12/22 14:59:22 JMI: poll_fast: returning -1 = GLOBUS_FAILURE (try Perl
> scripts)
>
> I am not sure what's going wrong,
Those messages are normal. What is the real problem you are having?
> can anyone give me a few words of wisdom th help here?
>
> and does anyone know if the code is documented somewhere,
> so I can see what does what etc without needing to wade through it?
> I mean the:
> Error.pm JobDescription.pm JobManager.pm JobState.pm submit-helper.pl
> Helper.pm JobManager JobSignal.pm StdioMerger.pm lcgpbs.pm
>
> cheers if u can help, it would be nice to get a job running before the
> hols start.
Those things are barely documented, I am afraid, and there are only a few
people that know the details. Let us start with a few basic things:
- What is the name of your CE?
- Are /opt/globus/tmp and /opt/globus/var/* world-writable and not full?
- Do you share home directories with your WNs?
|