 |
 |
Leslie Groer wrote: >In my testing of our newly configured site under LCG-2.3.1 with Condor as >the LRMS, it has become obvious that the default environment normally >setup by /etc/profile.d is not available on the worker nodes. > >The output from testJob.sh submitted with edg-job-submit is below. Note >the usual VO environmentals etc setup by lcgenv.sh are missing. The only >environmentals available are those passed in by the Condor batch system. > >Logging into the node as atlas005 sets up the environment correctly. I am >presuming this is an artefact of Condor job submission (as compared to >PBS, for instance). > >Will I need to force some sourcing of the /etc/profile.d/*.sh scripts or >can I assume all *real* job submissions will take care of this? Is this a >solved issue on other sites? > >Thanks >Leslie > > >+++++++++++++++++++++++++++++++++++++++++++++++++++ > node210 Wed Apr 6 14:44:11 EDT 2005 >+++++++++++++++++++++++++++++++++++++++++++++++++++ >the environment on the node > >EDG_WL_JOBID=https://lxn1188.cern.ch:9000/uDoGxKZwq6qv64w7La6IAw >EDG_WL_LOCATION=/opt/edg >EDG_WL_LOG_DESTINATION=bigmac-lcg-ce.physics.utoronto.ca >EDG_WL_RB_BROKERINFO=/condor/execute/dir_8900/globus-tmp.node210.8909.0/WMS_node210_09154_https_3a_2f_2flxn1188.cern.ch_3a9000_2fuDoGxKZwq6qv64w7La6IAw/.BrokerInfo >EDG_WL_SEQUENCE_CODE=UI=000003:NS=0000000003:WM=000004:BH=0000000000:JSS=000003:LM=000000:LRMS=000002:APP=000000 >GLOBUS_GASS_CACHE_DEFAULT=/condor/execute/dir_8900/globus-tmp.node210.8909.0/globus-tmp.node210.8909.0 >GLOBUS_GRAM_JOB_CONTACT=https://bigmac-lcg-ce.physics.utoronto.ca:20008/29766/1112812598/ >GLOBUS_GRAM_MYJOB_CONTACT=URLx-nexus://bigmac-lcg-ce.physics.utoronto.ca:20009/ >GLOBUS_LOCATION=/opt/globus >GLOBUS_PATH=/opt/globus >GLOBUS_REMOTE_IO_URL=/tmp/globus-tmp.node210.8909.2 >GLOBUS_TCP_PORT_RANGE=20000 25000 >HOME=/condor/execute/dir_8900/globus-tmp.node210.8909.0 >LD_LIBRARY_PATH=/opt/globus/lib >LIBPATH=/opt/globus/lib:/usr/lib:/lib >LOGNAME=atlas005 >PATH=/opt/globus/bin:/opt/globus/sbin:/usr/local/bin:/bin:/usr/bin >PWD=/condor/execute/dir_8900/globus-tmp.node210.8909.0/WMS_node210_09154_https_3a_2f_2flxn1188.cern.ch_3a9000_2fuDoGxKZwq6qv64w7La6IAw >SASL_PATH=/opt/globus/lib/sasl >SCRATCH_DIRECTORY=/home/atlas005/globus-tmp.node210.8909.0 >SHLIB_PATH=/opt/globus/lib >SHLVL=3 >X509_USER_PROXY=/tmp/globus-tmp.node210.8909.0 >_=/bin/env >_CONDOR_SCRATCH_DIR=/condor/execute/dir_8900 > >Leslie > > Do you use any job wrapper for condor ? We use in our local config file: USER_JOB_WRAPPER=/opt/condor-6.6.8/etc/condor_job_wrapper.sh which is simply: #!/bin/sh --login exec $* and it solves the problem. regards, Gergo