Back to Troubleshooting Guide
globus-job-run returns nothing
$ globus-job-run some-CE.domain/jobmanager-lcglsf -q queue /usr/bin/whoami $
The command returns nothing, it does not hang and exits with status 0.
This can happen for at least these reasons:
- The user has no permission to submit to the given queue. See Unspecified_gridmanager_error.
- The job manager is of "lcg" flavor (lcgcondor, lcglsf, lcgpbs, lcgsge) and the queue name is misspelled and it happens to be a substring of any existing queue name. For example, if the intended queue is "dteam", the command will return nothing if the queue argument is given as "tea".
- In /opt/globus/lib/perl/Globus/GRAM/JobManager/lcglsf.pm lines like these:
$bsub = '/usr/share/lsf/6.2/linux2.6-glibc2.3-x86/bin/bsub'; $bjobs = '/usr/share/lsf/6.2/linux2.6-glibc2.3-x86/bin/bjobs'; $bkill = '/usr/share/lsf/6.2/linux2.6-glibc2.3-x86/bin/bkill'; $bacct = '/usr/share/lsf/6.2/linux2.6-glibc2.3-x86/bin/bacct'; $bmod = '/usr/share/lsf/6.2/linux2.6-glibc2.3-x86/bin/bmod';
- may need to be replaced with lines like these:
my $LSF = ". /etc/profile.d/lsf.sh; " . "/usr/local/lsf/6.2/linux2.6-glibc2.3-x86/bin"; $bsub = "$LSF/bsub"; $bjobs = "$LSF/bjobs"; $bkill = "$LSF/bkill"; $bacct = "$LSF/bacct"; $bmod = "$LSF/bmod";
- to give the LSF commands the environment they need. Note that on an LCG-CE the globus-job-manager-marshal and globus-gma daemons need to be restarted when such changes are made. If the change fixes the problem, the same changes must be put in /opt/globus/setup/globus/lcglsf.in so that they will be preserved when the CE is reconfigured.