[gridengine users] Node with negative value of a consumable

William Hay w.hay at ucl.ac.uk
Tue Jun 14 10:58:36 UTC 2011


On 14 June 2011 11:38, Mark Dixon <m.c.dixon at leeds.ac.uk> wrote:
> On Tue, 14 Jun 2011, William Hay wrote:
> ...
>> Assuming this is for local scratch space on Linux one thing I've been
>> considering is using almost the whole disk as swap and then mounting a
>> suitably sized tmpfs on $TMPDIR in the prolog
>> followed by a umount in the epilog.  Have to be careful to ensure it
>> didn't cause real memory to be devoted to the VFS rather than
>> processes though.
> ...
>
> Sounds smarter than the lvm/mkfs scheme I had been thinking of. However...
>
> That's great for jobs that just need some fast local disk during the job,
> but it's a problem for people wanting to use it for data to be picked up
> after the job has finished (e.g. checkpoint files).
We have a mechanism in place for users to request their $TMPDIR be
saved to the shared filestore
by the epilog when the job has finished.  Doesn't help if you are
checkpointing to $TMPDIR to protect against
node crashes of course.
>
> Perhaps it's less of a problem these days, if you've got a fast parallel
> filesystem.

>
> TTFN
>
> Mark
> --
> -----------------------------------------------------------------
> Mark Dixon                       Email    : m.c.dixon at leeds.ac.uk
> HPC/Grid Systems Support         Tel (int): 35429
> Information Systems Services     Tel (ext): +44(0)113 343 5429
> University of Leeds, LS2 9JT, UK
> -----------------------------------------------------------------
>




More information about the users mailing list