[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [HTCondor-users] Disk space consumability

On 11/18/2019 5:32 PM, Oliver Freyermuth wrote:
> I *think* I also saw some jobs overcommitting storage, but that might be because our monitoring was broken.

Just a heads up... there are indeed a couple reasons why your dslot may 
end up with more Disk provisioned than requested by the job:

1. The default value for config knob MODIFY_REQUEST_EXPR_REQUESTDISK 
will round up a disk request:

  MODIFY_REQUEST_EXPR_REQUESTDISK = quantize(RequestDisk,{1024})

2. There is a long-standing limitation in the current implementation of 
the startd where Disk is rounded up to the nearest 1% of the total 
server Disk.  We will fix this in the code in the months to come.... 
while most users could care less today, as servers with more than 100 
cores become more common, people will start to care more. :)

> Many thanks for pointing me in the right direction and showing me HTCondor is doing exactly what I want :-).

Glad I was able to help!

best regards,