Re: [Condor-users] Condor-C to PBS

[Marc Ozonne <marc.ozonne@xxxxxxx>]

Hi all,
Thanks Francesco for your answer but there is something I still do not understand.
This is my configuration.
Host A
- condor daemons (schedd, collector, negotiator, startd)
- pbs_server, pbs_mom
- maui scheduler

Host B
- pbs_mom

This submission schema should work with Condor 6.8, shouldn't it ?
[ condor-c (+blahpd) --> pbs_server -->maui ]   -->   [pbs_mom]
                             Host A                                                Host B
If I submit a job through condor-c to the pbs_server, after contacting "maui", pbs_server should be able to send the job to HostB ? Am I wrong ?

Thanks for your help.
Cheers,
Marc


Francesco Prelz wrote:

What can I do to submit a condor-c job to the PBS queue ?
Do you have an example of submission file ?
    

There has to be a Condor schedd on Machine B, or a working PBS
client on machine A. The 'L' in BLAH stands for 'local', so the
submission will work only if the Condor daemons and the PBS client
live on the same host.

As you can easily sort out if you manage to find some of the EGEE people 
around CERN while they are not having meetings, in the EGEE context
the Condor daemons on 'B' are materialised by 'launching' them
via the Globus GRAM protocol and 'jobmanager-fork', as they have to
be running as the right user, that will eventually do the PBS submission.
Take a look at the scripts inside libexec/glite in the Condor
distribution if you need more details about how this works.
An example submit file for that setup is attached. It is taken
from a live system and I don't have time to clean it up, but that is
what we use in practice.

We are currently working on a different setup where one set of Condor 
daemons will be able to impersonate different users on B via 'glexec'.
This is not ready yet, but could be in a few weeks.

Francesco Prelz
INFN Milano
  

+REMOTE_SubUniverse = "blah"
requirements = Name=="1f6935e8b43657fdea0e18240e2919f9@xxxxxxxxxxxxxxxxxxxxx"
InitialDir = /var/glite/jobcontrol/condorio/RC/https_3a_2f_2fatlfarm001.mi.infn.it_3a9000_2fRCFAXR4aWujHCXqe3w9QUg/
stream_error = False
+edg_jobid = "https://atlfarm001.mi.infn.it:9000/RCFAXR4aWujHCXqe3w9QUg"
+REMOTE_JobUniverse = 9
+SiteName = "atlfarm006.mi.infn.it"
Arguments = 'UI=000000:NS=0000000007:WM=000014:BH=0000000000:JSS=000009:LM=000014:LRMS=000000:APP=000000'
GlobusScheduler = atlfarm006.mi.infn.it:2119/blah-pbs
+REMOTE_REMOTE_UserSubjectName = "/C=IT/O=INFN/OU=Personal Certificate/L=Milano/CN=Francesco Prelz/Email=francesco.prelz@xxxxxxxxxx"
+REMOTE_REMOTE_edg_jobid = "https://atlfarm001.mi.infn.it:9000/RCFAXR4aWujHCXqe3w9QUg"
+ce_id = "atlfarm006.mi.infn.it:2119/blah-pbs-infinite"
+SiteGatekeeper = "atlfarm006.mi.infn.it:2119/jobmanager-fork"
Output = /var/glite/jobcontrol/condorio/RC/https_3a_2f_2fatlfarm001.mi.infn.it_3a9000_2fRCFAXR4aWujHCXqe3w9QUg/StandardOutput
Submit_Event_Notes = (https://atlfarm001.mi.infn.it:9000/RCFAXR4aWujHCXqe3w9QUg) (UI=000000:NS=0000000007:WM=000014:BH=0000000000:JSS=000009:LM=000014:LRMS=000000:APP=000000) (0)
remote_schedd = $$(Name)
periodic_hold = Matched =!= TRUE && CurrentTime > QDate + 900
Copy_to_Spool = False
Executable = /var/glite/jobcontrol/submit/RC/JobWrapper.https_3a_2f_2fatlfarm001.mi.infn.it_3a9000_2fRCFAXR4aWujHCXqe3w9QUg.sh
+REMOTE_REMOTE_VirtualOrganisation = "EGEE"
+REMOTE_REMOTE_Stagecmd = True
+REMOTE_REMOTE_Queue = "infinite"
X509UserProxy = /var/glite/SandboxDir/RC/https_3a_2f_2fatlfarm001.mi.infn.it_3a9000_2fRCFAXR4aWujHCXqe3w9QUg/user.proxy
+REMOTE_JobGridType = "blah"
error = /var/glite/jobcontrol/condorio/RC/https_3a_2f_2fatlfarm001.mi.infn.it_3a9000_2fRCFAXR4aWujHCXqe3w9QUg/StandardError
+LB_sequence_code = "UI=000000:NS=0000000007:WM=000014:BH=0000000000:JSS=000009:LM=000014:LRMS=000000:APP=000000"
Notification = never
stream_output = False
GlobusRSL = (queue=infinite)(jobtype=single)(environment=(EDG_WL_HLR_LOCATION 'grid003.mi.infn.it:56568:') (EDG_WL_JOBID 'https://atlfarm001.mi.infn.it:9000/RCFAXR4aWujHCXqe3w9QUg'))
+REMOTE_REMOTE_ceid = "atlfarm006.mi.infn.it:2119/blah-pbs"
should_transfer_files = YES
+Type = "job"
Universe = grid
+UserSubjectName = "/C=IT/O=INFN/OU=Personal Certificate/L=Milano/CN=Francesco Prelz/Email=francesco.prelz@xxxxxxxxxx"
Log = /var/glite/logmonitor/CondorG.log/CondorG.1153935758.log
+REMOTE_REMOTE_VomsFqan = "/EGEE/Role=NULL/Capability=NULL"
grid_type = condor
transfer_input_files = /var/glite/SandboxDir/RC/https_3a_2f_2fatlfarm001.mi.infn.it_3a9000_2fRCFAXR4aWujHCXqe3w9QUg/user.proxy
when_to_transfer_output = ON_EXIT
+GlobusResource = "$$(Name)"
+CondorSubmitFile = "/var/glite/jobcontrol/submit/RC/Condor.https_3a_2f_2fatlfarm001.mi.infn.it_3a9000_2fRCFAXR4aWujHCXqe3w9QUg.submit"
+REMOTE_REMOTE_GridType = "pbs"
+REMOTE_Env = "X509_USER_PROXY=user.proxy;$$(GLITE_ENV:);EDG_WL_HLR_LOCATION=grid003.mi.infn.it:56568:"
+daemon_unique_name = "1f6935e8b43657fdea0e18240e2919f9"
Queue 1