Whats wrong with just using the tools as is?
Sid Young
On Thu, Sep 16, 2021 at 5:54 AM Ondrej Valousek
wrote:
> Hi list,
> I am wondering if there is a plugin allowing to submit jobs via SystemD
> (I.e. using systemd-run) on exec nodes.
>
> I have actually modified SGE sources to offer this o
Hi Machael,
Thanks for your quick response. All factors you mentioned look valid to me. For
1, a similar situation can also happen when there is a queued large job that
just failed.
I feel it can be bad if the backfill scheduler is enabled by default, but the
'-test-only' estimation doesn't con
One more reminder that the Slurm User Group Meeting (SLUG'21) will be
held on Tuesday, streaming through YouTube Live.
The agenda's been updated with the titles for each of the five sessions,
and links have been added to the individual streams:
https://slurm.schedmd.com/slurm_ug_agenda.html
I can imagine at least the following causing differences in the estimated time
and the actual start time:
* If running users have overestimated their job times, and their jobs
finish earlier than expected, the original estimate will be high.
* If another user's job submission gets highe
Hi and thanks for reading this!
I am trying to estimate the queue time of a job of a certain size and walltime
limit. I am doing this because our project considers multiple HPC resources and
needs estimated queue time information to decide where to actually submit the
job.
>From the man page o
Hi list,
I am wondering if there is a plugin allowing to submit jobs via SystemD
(I.e. using systemd-run) on exec nodes.
I have actually modified SGE sources to offer this option (available at
https://github.com/daimh/sge) and it works quite well.
The advantages are:
- we let systemd to track and
Hi,
How to bind group of users to a particular partition.
I have followed this :
Given allowgroup to partition in slurm.conf, but the users from the group
are able to submit jobs to other partitions.
Second option is :
Created QOS without any parameter, and given AllowQOS to
Hi all
Do I understand correctly that keyword
"PriorityType=priority/basic"
means that a job which is the first in queue will not let other jobs to
start before it? And this keyword is enough to switch on this mode?
The problem is following.
I have a job in queue which does not start long ti
