Maarten Litmaath wrote:
> Neither lcgpbs nor the pbs job manager care about EDG_WL_SCRATCH.
> That variable is taken into account only by the lcg-RB job wrapper:
Hello Maarten,
Thanks *very* much for this answer which really clarifies things.
I have been misled by the fact that WMS started to be used by both
ALICE (from the vobox) and OPS at about the same moment I changed the
jobmanager. I reproduced the problem by using RB submission and
EDG_WL_SCRATCH was actually used.
So I tried your suggestion using GLITE_LOCAL_CUSTOMIZATION_DIR
and had a SAM test successful using local storage at worker node.
I have a few questions and one suggestion to finish with and close
this :
a) is there a convention for the name of the temporary directory to
use (with the RB, the name was quite long and included a reference
to the job id http...) that could be useful to track a particuliar
job on the worker nodes ?
b) After job execution, the temporary dir is not suppressed and it
contains a '.rnd' file which I suppose is a seed for random
generator. Is this normal that the directory is not suppressed,
do I have to do sth for this ?
Suggestion : include code in the YAIM function config_wn like :
if [ "$GLITE_LOCAL_CUSTOMIZATION_DIR" ]; then
yaimgridenv_set GLITE_LOCAL_CUSTOMIZATION_DIR \
${GLITE_LOCAL_CUSTOMIZATION_DIR}
fi
This would add the variable definition in grid-env.zsh if people
want to use it...
Thanks again Maarten.
JM
--
------------------------------------------------------------------------
Jean-michel BARBET | Tel: +33 (0)2 51 85 84 86
Laboratoire SUBATECH Nantes France | Fax: +33 (0)2 51 85 84 79
CNRS-IN2P3/Ecole des Mines/Universite | E-Mail: [log in to unmask]
------------------------------------------------------------------------
|