Re: [HTCondor-users] shared fs high latency slow down the schedd

[Greg Thain <gthain@xxxxxxxxxxx>]

On 06/20/2017 10:43 AM, Krieger, Donald N. wrote:
> How about putting the SPOOL directory on /dev/shm ?  Do the files need to be preserved when the system is rebooted and /dev/shm/ gets cleared?
>
> Best - Don

SPOOL can get pretty big, as it contains job sandboxes. JOB_QUEUE_LOG 
(which defaults to a file under the SPOOL directory) points to the file 
which contains the schedd's transaction log, and is very frequently 
written to, and fsync'ed.  Putting JOB_QUEUE_LOG on /dev/shm mean the 
entire job queue will disappear when the machine reboots.

-greg