Re: [slurm-users] Running Slurm on a single host?

[Miguel Gutiérrez Páez]

Hi,

I have to say that I'm in this case. I have a node with quite resources and
shared by a few users. I've installed slurm to manage the resources of the
host properly. Yes, the users still can run jobs bypassing slurm, but for
now it is not necessary to implement cgroups or similar to reduce abuse.
So yes, I'm also use slurm (I have a whole slurm cluster runnig separately)
to manage the resources of a shared host.

Regards.

El vie., 6 abr. 2018 a las 20:43, stolarek.marcin (<
stolarek.mar...@gmail.com>) escribió:

> Maybe consol user can simply use interactive job? Other approach would be
> to reboot the computer to different os image, probably killing the job. You
> can also think about those hosts reservation for working hours to prevent
> long jobs start just before interactive use in form of slurm job or just
> someone using it directly even without server reboot..
>
> Cheers,
> marcin
>
>  Original message 
> From: Patrick Goetz 
> Date: 06/04/2018 06:00 (GMT-11:00)
> To: Slurm User Community List 
> Subject: [slurm-users] Running Slurm on a single host?
>
> I've been using Slurm on a traditional CPU compute cluster, but am now
> looking at a somewhat different issue.  We recently purchased a single
> machine with 10 high end graphics cards to be used for CUDA calculations
> and which will shared among a couple of different user groups.
>
> Does it make sense to use Slurm for scheduling in this case?  We'll want
> to do things like limit the number of GPU's any one user can use and
> manage resource contention the same way one would for a cluster.
> Potentially this would mean running slurmctld and slurmd on the same host?
>
> Bonus question: these research groups (they do roughly the same kind of
> work) also have a pool of GPU workstations they're going to share.  It
> would be super cool if we could somehow rope the workstations into the
> resource pool in cases where no one is working at the console. Because
> some of this stuff involves steps with interactive components, the
> understanding would be that all resources go to a console user when
> there is a console user.
>
>
>
>
>

Re: [slurm-users] Running Slurm on a single host?

[TO_Webmaster]

It might be possible to restrict all these ssh users into a very
limited cgroup and use the rest of the host for slurm jobs.

2018-04-06 19:30 GMT+02:00 Alex Chekholko :
> The thing you are describing is possible in both theory and practice. Plenty
> of people use a scheduler on a single large host.  The challenge will be in
> enforcing user practices so they don't just run commands directly but
> through the scheduler.
>
> On Fri, Apr 6, 2018 at 10:00 AM, Patrick Goetz 
> wrote:
>>
>> I've been using Slurm on a traditional CPU compute cluster, but am now
>> looking at a somewhat different issue.  We recently purchased a single
>> machine with 10 high end graphics cards to be used for CUDA calculations and
>> which will shared among a couple of different user groups.
>>
>> Does it make sense to use Slurm for scheduling in this case?  We'll want
>> to do things like limit the number of GPU's any one user can use and manage
>> resource contention the same way one would for a cluster. Potentially this
>> would mean running slurmctld and slurmd on the same host?
>>
>> Bonus question: these research groups (they do roughly the same kind of
>> work) also have a pool of GPU workstations they're going to share.  It would
>> be super cool if we could somehow rope the workstations into the resource
>> pool in cases where no one is working at the console. Because some of this
>> stuff involves steps with interactive components, the understanding would be
>> that all resources go to a console user when there is a console user.
>>
>>
>>
>>
>

Re: [slurm-users] Running Slurm on a single host?

[Alex Chekholko]

The thing you are describing is possible in both theory and practice.
Plenty of people use a scheduler on a single large host.  The challenge
will be in enforcing user practices so they don't just run commands
directly but through the scheduler.

On Fri, Apr 6, 2018 at 10:00 AM, Patrick Goetz 
wrote:

> I've been using Slurm on a traditional CPU compute cluster, but am now
> looking at a somewhat different issue.  We recently purchased a single
> machine with 10 high end graphics cards to be used for CUDA calculations
> and which will shared among a couple of different user groups.
>
> Does it make sense to use Slurm for scheduling in this case?  We'll want
> to do things like limit the number of GPU's any one user can use and manage
> resource contention the same way one would for a cluster. Potentially this
> would mean running slurmctld and slurmd on the same host?
>
> Bonus question: these research groups (they do roughly the same kind of
> work) also have a pool of GPU workstations they're going to share.  It
> would be super cool if we could somehow rope the workstations into the
> resource pool in cases where no one is working at the console. Because some
> of this stuff involves steps with interactive components, the understanding
> would be that all resources go to a console user when there is a console
> user.
>
>
>
>
>