[Condor-users] Unable to submit job to Globus (3.2) with Condor-G

[Tim Sheppard <t.sheppard@xxxxxxxxxxxxxxxxxx>]

Hi,

I have recently installed Condor (6.6.7) to use Condor-G as a job 
scheduler for an installation of Globus 3.2. I am able to submit the 
test jobs and they appear as idle in condor_q at first. However, the 
jobs never get submitted to globus to be run due to the following error:

From the job's log file:
000 (021.000.000) 11/04 10:47:18 Job submitted from host: 
<192.168.168.20:32772>
...
018 (021.000.000) 11/04 10:47:21 Globus job submission failed!
   Reason: 7 an authentication operation failed
...
012 (021.000.000) 11/04 10:47:26 Job was held.
   Globus error 7: an authentication operation failed
   Code 2 Subcode 7

and the from the condor log file:
11/4 10:47:18 ******************************************************
11/4 10:47:18 ** condor_gridmanager (CONDOR_GRIDMANAGER) STARTING UP
11/4 10:47:18 ** /usr/local/condor/sbin/condor_gridmanager
11/4 10:47:18 ** $CondorVersion: 6.6.7 Oct 11 2004 $
11/4 10:47:18 ** $CondorPlatform: I386-LINUX_RH9 $
11/4 10:47:18 ** PID = 20587
11/4 10:47:18 ******************************************************
11/4 10:47:18 Using config file: /etc/condor/condor_config
11/4 10:47:18 Using local config files: 
/usr/local/condor/local.snodhill/condor_config.local
11/4 10:47:18 DaemonCore: Command Socket at <192.168.168.20:39821>
11/4 10:47:18 [20587] GAHP server pid = 20588
11/4 10:47:21 [20587] DaemonCore: Command received via UDP from host 
<192.168.168.20:32791>
11/4 10:47:21 [20587] DaemonCore: received command 60000 
(DC_RAISESIGNAL), calling handler (HandleSigCommand())
11/4 10:47:21 [20587] Found job 21.0 --- inserting
11/4 10:47:21 [20587] (21.0) doEvaluateState called: gmState GM_INIT, 
globusState 32
11/4 10:47:21 [20587] (21.0) proxy not cached yet, waiting...
11/4 10:47:21 [20587] (21.0) doEvaluateState called: gmState GM_INIT, 
globusState 32
11/4 10:47:21 [20587] resource 192.168.168.20:8081 is now up
11/4 10:47:21 [20587] (21.0) doEvaluateState called: gmState GM_INIT, 
globusState 32
11/4 10:47:21 [20587] (21.0) doEvaluateState called: gmState GM_SUBMIT, 
globusState 32
11/4 10:47:21 [20587] (21.0) gmState GM_SUBMIT, globusState 32: 
globus_gram_client_job_request() returned Globus error 7
11/4 10:47:21 [20587] (21.0)    
RSL='&(rsl_substitution=(GRIDMANAGER_GASS_URL 
https://snodhill.dtg.qinetiq.com:39828))(executable='/bin/ls')(scratchdir='')(directory=$(SCRATCH_DIRECTORY))(stdout=$(GRIDMANAGER_GASS_URL)#'/home/tim/ls-test.out')(proxy_timeout=240)(save_state=yes)(two_phase=600)(remote_io_url=$(GRIDMANAGER_GASS_URL))'
11/4 10:47:26 [20587] No jobs left, shutting down
11/4 10:47:26 [20587] Got SIGTERM. Performing graceful shutdown.
11/4 10:47:26 [20587] **** condor_gridmanager (condor_GRIDMANAGER) 
EXITING WITH STATUS 0

I have already found an article that explains that Condor (and Condor-G) 
use an older version of openssl to Globus 3.2 and I therefore create a 
legacy proxy before submitting the job (by using the -old option) but it 
seems I am still unable to authenticate Condor-G with Globus to allow 
the job submission.

Is anyone able to help me with this?

Many Thanks,

Tim