[HTCondor-users] GPUs_MONITOR resource usage

[Alexander Antoniades <sander@xxxxxxxxxxxx>]

Hi,

I have a consumer gpu server from Advanced HPC which has 8 1080ti GPUs and 2ÂE5-2603 Xeons (6 cores each).

The cpu usage of GPUs_MONITOR seems a little high for sitting at idle.

condor  2165 66.9 0.0 26460 6844 ?    R  Jan08 793:42 GPUs_MONITOR

This is with Condor 8.8 onÂUbuntu 18.04.1 LTS.

This is the relevent portion of the node configuration:

DAEMON_LIST = MASTER,STARTD

## Always run jobs

START = True

SUSPEND = False

CONTINUE = True

PREEMPT = False

KILL = False

## GPU Discovery per https://htcondor-wiki.cs.wisc.edu/index.cgi/wiki?p=HowToManageGpus

use feature : GPUs

#GPU_DISCOVERY_EXTRA = -extra

This is a pretty vanilla configuration we are using for testing, this should be the relevant portion of the master node:

DAEMON_LIST = MASTER, COLLECTOR, NEGOTIATOR, SCHEDD

NEGOTIATOR_INTERVAL = 30

NEGOTIATE_ALL_JOBS_IN_CLUSTER = True

Is the resource usage of the monitoring process normal?

Thanks,

Sander