Hi Kashif,
> I have installed a lcg-CE fronting a tarball WN installation and using
> PBS as JOB_MANAGER. It is also using shared home directory of pool
> accounts between CE and WN. I am submitting job using -r option. Job is
> successfully submitting to CE and I can see from batch system that it
> get completed.
> But when I check glite-wms-job-status, it keep showing 'running' until
> proxy get expired.
> This bit keeps repeating in /home/user/gram_job_mgr_*
>
> 10/1 18:00:47 JMI: poll: seeking:
> https://ngsce-test.oerc.ox.ac.uk:64001/1937/12
> 54416006/
> 10/1 18:00:47 JMI: poll_fast: ******** Failed to find
> https://ngsce-test.oerc.ox
> .ac.uk/1937/1254416006/
> 10/1 18:00:47 JMI: poll_fast: returning -1 = GLOBUS_FAILURE (try Perl
> scripts)
> 10/1 18:00:47 JMI: cmd = poll
> 10/1 18:00:47 JMI: returning with success
Those errors look "normal".
> I have already checked that I can do globus-url-copy from WN pool
> account to CE as well WMS.
With the standard jobmanager-pbs your WNs do not need access to the CE!
> Crl are up to date
> There is no issue of ssh key as home directory is shared.
> I am using GLOBUS_TCP_PORT_RANGE="64000,65256" for some reason.
Check here:
http://goc.grid.sinica.edu.tw/gocwiki/Jobs_sent_to_some_CE_stay_in_Running_state_forever
That page redirects here:
http://goc.grid.sinica.edu.tw/gocwiki/Jobs_sent_to_some_CE_stay_in_Scheduled_state_forever
|