I'm running condor 6.8.2 on solaris platforms. When I submit a condor_dagman job like:
condor_submit_dag -verbose -force -notification never -outfile_dir /tmp -usedagdir /tmp/condor9198.dag
where the DAG has one job in it, the resultant condor_dagman job is scheduled and in the 'R' state but the job in the dag remains in the 'I' for roughly 5 minutes before executing.
After this 5 minute delay the job runs successfully. Also, when scheduling the job without a DAG, it runs immediately and also successfully.
Probably not coincidently, there are a few timeout configuration settings set at 300 seconds.
Anyone seen this behaviour ?
Northrop Grumman Corporation
Baltimore, MD 21203