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

[Todd Tannenbaum <tannenba@xxxxxxxxxxx>]

Another idea is put
  CONDOR_FSYNC = False
in your condor_config, as fsync is often the performance bottleneck on 
shared filesystems.  Of course, in the event of a system crash, it is 
possible that jobs could "disappear" from the queue if they were 
recently submitted, but maybe you can live with that. From the HTCondor 
Manual:

CONDOR_FSYNC
    A boolean value that controls whether HTCondor calls fsync() when 
writing the user job and transaction logs. Setting this value to False 
will disable calls to fsync(), which can help performance for 
condor_schedd log writes at the cost of some durability of the log 
contents, should there be a power or hardware failure. The default value 
is True.


On 6/20/2017 12:58 PM, Dimitri Maziuk wrote:
> On 06/20/2017 10:43 AM, Krieger, Donald N. wrote:
> > How about putting the SPOOL directory on /dev/shm ?
>
> Aside from the disappearing job queue, filling up /dev/shm may cause all
> kinds of interesting effects, depending on how and where you mount it.
>
> This is something to keep in mind when moving spool to local drive, too:
> we recently filled up / and had to move spool to a zfs pool -- the part
> where *spool can get large* should be in all-caps bold in the manual.
>
>
>
> _______________________________________________
> HTCondor-users mailing list
> To unsubscribe, send a message to htcondor-users-request@xxxxxxxxxxx with a
> subject: Unsubscribe
> You can also unsubscribe by visiting
> https://lists.cs.wisc.edu/mailman/listinfo/htcondor-users
>
> The archives can be found at:
> https://lists.cs.wisc.edu/archive/htcondor-users/


--
Todd Tannenbaum <tannenba@xxxxxxxxxxx> University of Wisconsin-Madison
Center for High Throughput Computing   Department of Computer Sciences
HTCondor Technical Lead                1210 W. Dayton St. Rm #4257
Phone: (608) 263-7132                  Madison, WI 53706-1685