Re: [HTCondor-users] job folders left on cluster

[Todd Tannenbaum <tannenba@xxxxxxxxxxx>]

On 3/23/2018 11:32 AM, Ryter, Derek wrote:
> This doesn't happen consistently, but at times a server will be 
> completely filled with folders created by jobs in the condor\execute 
> folder. I'm running pest++. Is it because of how a run terminates, or 
> something missing in the submit file? Any ideas? 

What version of HTCondor are you using?

I ask because this sounds to me like it could be a bug that got fixed in 
HTCondor v8.6.9 and above.

In the Version History for v8.6 (at http://htcondor.org/manual, then 
click on 'Version History'), note the following entry:

"Fixed a bug on Windows that would often result in the job sandbox on 
the execute node not being deleted when the condor_schedd relinquished 
its claim on the slot before the condor_starter had exited. ( Ticket 
#6497 - https://condor-wiki.cs.wisc.edu/index.cgi/tktview?tn=6497 )"

Hope the above helps,
regards,
Todd