Re: [HTCondor-users] Job matching in mixed architecture clusters? (x86_64 and AArch64)

[Ben Jones <ben.dylan.jones@xxxxxxx>]

Hi,

For grid we have a transform to explicitly set a Requirement for Arch to make sure the right jobs end up on the right machines. For a specific requirements (ie ARM) we normally have VOs set an attr or use the remote_queue thingy, so we end up with:

JOB_ROUTER_TRANSFORM_x86 @=end
    NAME x86
    REQUIREMENTS (MY.WantARM =!= True && MY.queue =!= "ARM")
    COPY Requirements PresArchRequirements
    SET Requirements (Arch =?= "X86_64" && PreArchRequirements)
@end

JOB_ROUTER_TRANSFORM_Arm @=end
    NAME Arm
    REQUIREMENTS (MY.WantARM =?= True) || (MY.queue =?= "ARM")
    SET WantARM True
    COPY Requirements PresArchRequirements
    SET Requirements (Arch =?= "aarch64" && PreArchRequirements)
@end

Not sure if this is âcorrectâ but it works for us.

cheers,
Ben

> On 27 May 2024, at 15:10, Fischer, Max (SCC) <max.fischer@xxxxxxx> wrote:
> 
> Hi all,
> 
> weâre almost ready to switch on some new toys in our cluster, a set of ARM servers - and just realised we have no idea whatâs âtheâ way of handling mixed x86_64 and AArch64 resources and jobs. Of course we can (and didâ *cough*) hack together our own rules, but is there a standard way for this?
> 
> The docs on job `Requirements` say âThe schedd appends a similar clause to match the Arch.â but a) I donât actually see any such rule on existing jobs and b) our schedds all are x86_64 with remote submissions (HTC-CE) so their arch is inconsequential.
> 
> Cheers,
> Max
> 
> _______________________________________________
> 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/