Hm, I'd need a container log in debug mode i think.

Define

   log4j.category.org.globus=DEBUG

in

   $GLOBUS_LOCATION/container-log4j.properties

and restart the GT4 container, repeat the condor-g submission
and send the GT4 logfile.

-Martin

Jaime Frey wrote:
> Yesterday, I was helping Klaus Schwarzmeier (who's subscribed to this
> list) with some problems he's having with his Globus 4.2.1 installation
> at Condor Week. He's trying to set up WS GRAM submit point into his
> Condor pool.
> 
> The Fork jobmanager is working fine, but the Condor jobmanager gets
> stuck in StageIn state. The odd part is that the job get submitted into
> the Condor pool, runs, and completes successfully. But 15 minutes later,
> WS GRAM still reports the job's status as StageIn. The client is
> Condor-G, which is receiving no job status notifications after one for
> StageIn, and active queries of the job's status return StageIn. The
> container log has an entry acknowledging the job submission to Condor,
> including the correct Condor job id.
> 
> We ran out of time before I could think of any further debugs steps to try.
> 
> +--------------------------------+-----------------------------------+
> |           Jaime Frey           | I used to be a heavy gambler.     |
> |       [email protected]        | But now I just make mental bets.  |
> |                                | That's how I lost my mind.        |
> +--------------------------------+-----------------------------------+
> 

Reply via email to