Re: [HTCondor-users] Prioritizing machines for match making

[Jaime Frey <jfrey@xxxxxxxxxxx>]

Do your jobs have a Rank _expression_? That will take priority in this NEGOTIATOR_PRE_JOB_RANK and wonât be reflected in your condor_status query.

 - Jaime

On Feb 27, 2024, at 12:01âAM, Vikrant Aggarwal <ervikrant06@xxxxxxxxx> wrote:

Hello Experts,

We are using scheduler level splitting for job match making.

Objective: We want to steer the job towards more busy machines. 

Setup Details: 

We are dynamically spinning up the nodes in the cloud based on the number of jobs in the queue. Once nodes are up, we keep on sending more jobs as the existing batch completes. We want to pack new jobs on machines where fewer cores are available so that one first batch with more jobs complete so that we can kill the vacant machines.. 

condor_config_val NEGOTIATOR_PRE_JOB_RANK (10000000 * My.Rank) + (1000000 * (RemoteOwner =?= UNDEFINED)) - (100000 * Cpus) - Memory

Jobs are randomly getting allocated to the nodes. Ex: I was expecting a node with 372604 NEGOTIATOR_PRE_JOB_RANK value would be preferred to run the jobs but it's picking the node with value in negative. 

[2024-02-23 07:08:49 root@xxxxxxxxxxxxxxxxxxxxxxxxxxx ~]# condor_status -compact -af '(10000000 * My.Rank) + (1000000 * (RemoteOwner =?= UNDEFINED)) - (100000 * Cpus) - Memory' machine rank remoteowner cpus memory | sort -r -n  -k1,1 | head
372604.0 c7condor1-default-us-c1-qb20.c.test.internal 0.0 undefined 6 27396
-147876.0 c7condor1-default-us-c1-zkz5.c.test.internal 0.0 undefined 11 47876

How can I force the behavior where nodes with less number of available cpus preferred? 

Thanks & Regards,

Vikrant Aggarwal

_______________________________________________
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/