Dear All...
I was about to submit a patch for the CREAMCE / SGE integration when I
hit the following issue...
---*---
1./ I can perfectly submit a job using glite-ce-job-submit . Check the
following examples:
-bash-3.00$ glite-ce-job-submit -a -r
ce03.lip.pt:8443/cream-sge-dteamgrid sleep.jdl
2009-10-13 16:17:31,114 WARN - No configuration file suitable for
loading. Using built-in configuration
https://ce03.lip.pt:8443/CREAM469661735
-bash-3.00$ glite-ce-job-status https://ce03.lip.pt:8443/CREAM469661735
2009-10-13 16:18:42,920 WARN - No configuration file suitable for
loading. Using built-in configuration
****** JobID=[https://ce03.lip.pt:8443/CREAM469661735]
Status = [DONE-OK]
ExitCode = [0]
---*---
2./ However, trying to submit via WMS, using
glite-wms-job-submit -a -r ce03.lip.pt:8443/cream-sge-dteamgrid
sleep.jdl,
the job never passes from the READY state.
---*---
3./ Monitoring the messages, glite-ce-cream and glite-ce-monitor logs, I
conclude that the job never reaches my SGE CREAMCE.
---*---
4./ I checked that the WMS workload manager registered the job,
13 Oct, 16:30:17 -I: [Info] operator()(dispatcher_utils.cpp:218):
new jobsubmit for https://wms01.lip.pt:9000/ctA8jPXMMyayYdOUMs84cA
13 Oct, 16:30:17 -I: [Info] operator()(submit_request.cpp:478):
https://wms01.lip.pt:9000/ctA8jPXMMyayYdOUMs84cA delivered
but I can not obtain its condor ID.
[root@wms01 CondorG.log]# grep
https://wms01.lip.pt:9000/_yUTkiOYPrcAJbRpjOGOdw
/var/glite/logmonitor/CondorG.log/*
[root@wms01 CondorG.log]#
---*---
5./ Finally, when I cancel the job, I see in jobcontoller_events.log,
the following log:
13 Oct, 16:44:09 -I- ControllerLoop::run(): Got new remove request (JOB
ID = https://wms01.lip.pt:9000/_yUTkiOYPrcAJbRpjOGOdw)...
13 Oct, 16:44:09 -I- JobControllerReal::cancel(...): Asked to remove
job: https://wms01.lip.pt:9000/_yUTkiOYPrcAJbRpjOGOdw
13 Oct, 16:44:09 -M- JobControllerReal::readRepository(): Reading
repository from LogMonitor file:
/var/glite/logmonitor/internal/irepository.dat
13 Oct, 16:44:10 -*- JobControllerReal::cancel(...): I'm not able to
retrieve the condor ID.
---*---
6./ I've tried to submit to other CREAM-CEs, bit I've noticed that a
list-match do not show me any cream-ce
-bash-3.00$ glite-wms-job-list-match -a sleep.jdl | grep -i cream
-bash-3.00$
My JDL is quite simple, and I do not think that I should have problems
with it:
-bash-3.00$ cat sleep.jdl
Executable = "sleep.sh";
StdOutput = "sleep.out";
StdError = "sleep.err";
RetryCount = 0;
InputSandbox = {"sleep.sh"};
OutputSandbox = {"sleep.out","sleep.err"};
#OutputSandboxBaseDestUri="gsiftp://ce02.lip.pt/tmp";
I've tried to force other creamCEs, using the "-r" option, but I got the
same results.
I conclude that something must be wrong in my WMS / ICE configuration...
---*---
7./ I'm using the following software:
[root@wms01 ~]# rpm -qa | grep glite-WMS
root@ce03 ~]# rpm -qa | grep CREAM
glite-CREAM-3.1.20-0
Any help is welcome...
Thanks in Advance
Cheers
Goncalo
