[HTCondor-users] Inconsistent JobCurrentStartDate and Event log?

[Kevin Retzke <kretzke@xxxxxxxx>]

We use JobCurrentStartDate to determine how long a job has been running, for enforcing runtime limits and CPU efficiency.

Lately, however, we've noticed that jobs can have up to several hours difference between JobCurrentStartDate and 

the Execute event in the EventLog. Examples:

$ condor_q 20628110.0 -af jobcurrentstartdate

1496406832

$ date -r 1496406832

Fri Jun  2 07:33:52 CDT 2017

$ grep 20628110.0 /var/log/condor/EventLog

000 (20628110.000.000) 06/02 07:32:17 Job submitted from host: ...

028 (20628110.000.000) 06/02 07:32:17 Job ad information event triggered.

001 (20628110.000.000) 06/02 10:48:57 Job executing on host: ...

$ condor_q 20582231.3250 -af jobcurrentstartdate

1496417203

$ date -r 1496417203

Fri Jun  2 10:26:43 CDT 2017

$ grep 20582231.3250 /var/log/condor/EventLog

000 (20582231.3250.000) 05/31 22:42:36 Job submitted from host: ...

028 (20582231.3250.000) 05/31 22:42:36 Job ad information event triggered.

001 (20582231.3250.000) 06/02 11:34:48 Job executing on host: ...

The event log appears to be correct, based on other job and file transfer logs. JobCurrentStartDate appears to always be earlier.

Anyone know why this might be?

Thanks,

Kevin