Re: [HTCondor-users] How do distribution jobs evenly on disks on the worknode

[Cole Bollig <cabollig@xxxxxxxx>]

Hi Zhangqier,

I think the best solution would be on the OS layer outside of HTCondor. You could create a raid 0 striped volume with the two disks to logically make them one with writes alternating between them. Thus, distributing the load evenly between the two physical disks for all jobs running on the execution point (worker node).

Cheers,

Cole Bollig

---

From: HTCondor-users <htcondor-users-bounces@xxxxxxxxxxx> on behalf of Qier W <wangzqe@xxxxxxxxx>
Sent: Sunday, October 1, 2023 7:05 AM
To: HTCondor-Users Mail List <htcondor-users@xxxxxxxxxxx>
Subject: [HTCondor-users] How do distribution jobs evenly on disks on the worknode

 

Dear experts,

   We have a worknode with a lot CPUs (couple hundreds) and we have two disks on the machine. How can we configure so that we can split condor jobs so that 1/2 of them uses one disk and another 1/2 uses another? Because we are afraid that a disk may not be able to handle this many jobs.

Sincerely,
Zhangqier Wang