Re: [Condor-users] state file location in CondorG

[Jaime Frey <jfrey@xxxxxxxxxxx>]

On Jun 2, 2005, at 11:25 AM, Erik Paulson wrote:

On Thu, Jun 02, 2005 at 03:54:11PM +0100, Dr Ian C. Smith wrote:

I'm trying to get CondorG going but when I submit

a job it keeps going into the held (H) state. The

GRAM logfile revealed this error:

6/2 15:36:32 JM: Creating and locking state lock file

6/2 15:36:32 JM: Failed to open state lock file

'/tmp/globus/job.ulgsmp1.liv.ac.uk.29536.1117722989.lock', e rrno=2

6/2 15:36:32 JM: error writing the state file

When I created /tmp/globus on the remote machine with 777 permission

it seems to work file. globus-job-run seems to work OK without

this so I'm assuming that its location is specified in Condor ?

No, it's specified in the job manager config file for Globus, it

has nothing to do with Condor. I think it's from

$GLOBUS_LOCATION/etc/globus-job-manager.conf

Correct.

The reason why globus-job-run works is because it invokes the GRAM jobmanager in a non-recoverable mode with no persistent state file. If the jobmanager on the server crashes, you have no way to recover. Your job is lost in the ether, still running.

+----------------------------------+---------------------------------+

|            Jaime Frey            |  Public Split on Whether        |

|        jfrey@xxxxxxxxxxx         |  Bush Is a Divider              |

|  http://www.cs.wisc.edu/~jfrey/  |         -- CNN Scrolling Banner |

+----------------------------------+---------------------------------+