Ciao Alessandro,
> since our CE has been migrated to sl4 and gLite3.1 (WNs already migrated
> to gLite 3.1 since some months) there is a job submission problem: it
> seems that every job submitted never arrives on the CE (the SAM tests
> are failing for "proxy expired"), the command
> $ globus-job-run gridit-ce-001.cnaf.infn.it/jobmanager-lcgpbs -queue
> cert /bin/pwd
>
> returns nothing, but ssh from WNs to CE and qsub are working
>
> Do you have any hints? I'm also attaching the file gram_job_mgr_24157
> related to my last "globus-job-run" attempt
In other gram_job_mgr_*.log files there are complaints like this:
-------------------------------------------------------------------------
Thu May 1 04:31:25 2008 JM_SCRIPT: executable staging failed with Cannot
connect socket to /opt/globus/var/globus-gass-cache-marshal.sock:
No such file or directory
-------------------------------------------------------------------------
Indeed:
-------------------------------------------------------------------------
$ globus-job-run gridit-ce-001.cnaf.infn.it /bin/ps auxwww |
grep 'marshal: accepting'
root 3369 0.0 0.0 12920 8240 ? Ss Apr29 0:06
globus-job-manager-marshal: accepting connections
-------------------------------------------------------------------------
Compare:
-------------------------------------------------------------------------
$ globus-job-run ce101.cern.ch /bin/ps auxwww | grep 'marshal: accepting'
root 13249 0.4 0.5 80500 12204 ? Ss Apr29 11:33
globus-job-manager-marshal: accepting connections
root 13251 0.0 0.2 66920 4360 ? Ss Apr29 1:08
globus-gass-cache-marshal: accepting connections
-------------------------------------------------------------------------
Try this:
/etc/init.d/globus-gass-cache-marshal restart
|