Re: [HTCondor-users] HTCondor Disk Benchmarks

["Bockelman, Brian" <BBockelman@xxxxxxxxxxxxx>]

Hi Alex,

I scoured a few of my configurations - couldn't find any immediate examples of my personal use of BENCHMARKS_JOBLIST.  When we've done benchmarks, we've done this prior to HTCondor's startup and encoded the results in the ad (this is a side-effect of the particular use case as we restart HTCondor daily ... it's a bit of an odd one).

That said, if I wanted to tackle the same problem today, BENCHMARKS_JOBLIST is likely how I'd go about it.

Note that the RANK mechanism is somewhat tricky to use: if your cluster is always full, then there are often few idle slots to chose between -- hence not many possible idle machines to sort through.  YMMV - it's definitely useful if the cluster ramps up / down a lot.

Brian

> On Mar 22, 2020, at 2:37 PM, Alex Waite <alex@xxxxxxxx> wrote:
> 
> Hello Everyone,
> 
> I would like jobs to be able to rank machines in my pool based off of the performance of the disk mounted at /var/lib/condor/execute/.
> 
> I'm not aware of any such functionality builtin to HTCondor, but I see two approaches:
> 1) configure nodes with static classads such as "fast_scratch"
> 2) extending the benchmarking process to test the disk
> 
> Going the benchmarking route is more appealing to me, and I found the BENCHMARKS_JOBLIST[1], so it should be possible.
> 
> Has anyone here attempting something similar (or at least used BENCHMARKS_JOBLIST) and would be willing to share their experience and/or config?
> 
> Thanks.
> 
> ---Alex
> _______________________________________________
> 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/