Hi,
Am 30.06.2014 um 08:55 schrieb Derrick Lin:
> A typical node on our cluster has 64 cores and 512GB memory. So it's about
> 8GB/core. Occasionally, we have some jobs that utilizes only 1 core but
> 400-500GB of memory, that annoys lots of users.
This is a general question, how you judge a nodes as empty:
- A job uses only one core but all available memory on a node. Is the node
fully loaded or not?
- A job uses all cores but almost no memory on a node. Is the node fully loaded
or not?
When a job has these requirements it's just the way it is, as other say too.
Do you want to "fool" other users that the heavy memory job of another user is
also using all cores (when they check the load they will spot it anyway), so
they think the usage is "balanced"?
> So I am seeking a way that can force jobs to run strictly below 8GB/core
> ration or it should be killed.
>
> For example, the above job should ask for 64 cores in order to use 500GB of
> memory (we have user quota for slots).
>
> I have been trying to play around h_vmem, set it to consumable and configure
> RQS
>
> {
> name max_user_vmem
> enabled true
> description "Each user can utilize more than 8GB/slot"
> limit users {bad_user} to h_vmem=8g
> }
Was "h_vmem" made consumable or is it just a limit?
-- Reuti
> but it seems to be setting a total vmem bad_user can use per job.
>
> I would love to set it on users instead of queue or hosts because we have
> applications that utilize the same set of nodes and app should be unlimited.
>
> Thanks
> Derrick
> _______________________________________________
> users mailing list
> users@gridengine.org
> https://gridengine.org/mailman/listinfo/users
_______________________________________________
users mailing list
users@gridengine.org
https://gridengine.org/mailman/listinfo/users
