On 9/20/21 6:10 PM, Marco Mambelli wrote:
Hi Greg, we collect job stats from the StarterLog: how many jobs started, completed, if they ended w/ exit code 0, not 0 or because of a signal. Is it there a better way to get this?
The startd keeps a startd_history file on the worker node that looks like a standard condor_history file, and can be read with the condor_history -file path-to-startd-history. I think this has all the info you need, and can generate the data in an easy to parse form.
-greg