So the way we handle it is that we give a blanket fairshare to everyone
but then dial in our TRES charge back on a per partition basis based on
hardware. Our fairshare doc has a fuller explanation:
https://docs.rc.fas.harvard.edu/kb/fairshare/
-Paul Edmon-
On 9/17/2020 9:30 AM, Mark Dixon wrote:
Hi all,
Clusters sometimes have a couple of different types of hardware, e.g.
lots of standard plus small amounts of highmem - with a partition per
type.
Sometimes one partition, e.g. "standard", is much busier than e.g.
"highmem".
In a fair share set up with multiple accounts and multiple users per
account, a single user making heavy use of an otherwise idle partition
can negatively effect their account's priority on the busy partition.
So having fair share per partition can be useful.
I imagine I can handle this by, for each group of users I'd normally
create a single account for, creating a duplicate "highmem" account. I
could then use a job_submit script to change the job's account to its
buddy "highmem" account when it sees jobs submitted to the himem
partition.
But this is a bit of a faff for reporting and managing users,
particularly as I've not figured out how to define a group of users in
one place (say, a unix group) that I can then use multiple times.
Is there a better way, please?
Thanks,
Mark
