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

Re: [HTCondor-users] Try to set DedicatedScheduler = NO JOBS EVER onto WN :[



Hi Winnie,

Jobs routed by an HTCondor-CE will have "RoutedBy = htcondor-ce" in 
their ClassAd so you could set the following configuration on your 
worker nodes that should only run CE jobs:

START = RoutedBy == "htcondor-ce"

As for your second question about submitting jobs to your CE requesting 
a specific node, the default configuration clears an incoming job's 
Requirements expression. The idea is that remote submitters would have 
to know intimate details about your local site policy and pool makeup to 
write any sensible job requirements.

However, we do have some documentation on how to preserve the original 
job's "Requirements" expression in your CE configuration [1]. After 
that's applied and you run `condor_ce_reconfig`, you could add the 
following to your submit file:

Requirements = Machine == "<WN hostname>"

Which would result in a routed job that requests a specific worker node. 
>From there, you can use `condor_q -better` and `condor_q -better 
-reverse` [2] for some ideas on why jobs and machines aren't matching.

- Brian

[1] 
https://htcondor-ce.readthedocs.io/en/latest/batch-system-integration/#preserving-original-job-requirements
[2] https://htcondor.readthedocs.io/en/latest/man-pages/condor_q.html

On 1/8/20 5:16 AM, Winnie Lacesso wrote:
> Greetings friendly helpful htcondor folk,
>
> Humbly, I've little KnowHow about htcondor. We have a CE (lcgce01) that
> accepts WLCG Grid jobs, several local-user-submit nodes, a "manager /
> matchmaker" htc00, & lotsa WN.
>
> On some of the WN we want to restrict them to only running Grid jobs, not
> local-user jobs. Experimenting on one of these WN, it seems when we try to
> set it to tell the "manager / matchmaker" "I only accept jobs from
> lcgce01", it then apparently gets NO jobs, ever. :[
>
> The advice from our local expert (who admits he doesn't know quite how to
> do this) is to setup the DedicatedScheduler on the WN.
>
> Copying from the example on the WN, this is in the WN's
> /etc/condor/config.d/20_workernode.config
>
> DedicatedScheduler = "DedicatedScheduler@xxxxxxxxxxxxxxxxxxxxxx"
> ##--------------------------------------------------------------------
> ## 1) Only run dedicated jobs
> ##--------------------------------------------------------------------
> START          = Scheduler =?= $(DedicatedScheduler)
> SUSPEND        = False
> CONTINUE       = True
> PREEMPT        = False
> KILL           = False
> WANT_SUSPEND   = False
> WANT_VACATE    = False
> RANK           = Scheduler =?= $(DedicatedScheduler)
>
> Is there a syntax error or something?
>
> Testing this would be easier if I knew how to submit a job to lcgce01
> saying "run this on the named WN" (then the matchmaker htc00 might log
> some error?)
> - but that seems impossible? I could not find where in the documentation
> this might be shown HowTo.
> If it is possible, what is the syntax to put in a test.xrsl submit file?
> (Or a pointer to documentation is welcome!)
>
> Humble very-non-expert!
>
>
> _______________________________________________
> 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/