Hi Carles, Carles Fenoy <[email protected]> writes:
> Re: [slurm-dev] Re: NumNodes changed for requeued job? > > Hi Loris, > > AFAIK for a job to be able to use more nodes than requested it should be: > > #SBATCH -N5- > > note the dash at the end. That specifies that the minimum number of > nodes is 5 and there is no maximum Thanks for the information. In our version of the man page for 2.2.7 it says ,--------------------------------------------------------------------- | -N, --nodes=<minnodes[-maxnodes]> | Request that a minimum of minnodes nodes be allocated to this | job. The scheduler may decide to launch the job on more than | minnodes nodes. `--------------------------------------------------------------------- I see that on the website that this now reads ,------------------------------------------------------------------------- | -N, --nodes=<minnodes[-maxnodes]> | Request that a minimum of minnodes nodes be allocated to this job. A | maximum node count may also be specified with maxnodes. If only one | number is specified, this is used as both the minimum and maximum | node count. `------------------------------------------------------------------------- So this bug in the documentation has obviously been fixed. Regards Loris > regards, > Carles Fenoy > > On Fri, Oct 19, 2012 at 11:11 AM, Loris Bennett > <[email protected]> wrote: > > Seren Soner <[email protected]> writes: > > > Re: [slurm-dev] NumNodes changed for requeued job? > > > > > You can normally request a range of number of nodes, i.e. > > > > #SBATCH -N 5-10 > > > > Requesting an exact number of nodes is equivalent to setting the > > minimum > and > > maximum number of nodes requested. After the job is scheduled, > > scontrol > show jo > > will convert the NumNodes field to the exact number of nodes > > that the job > uses. > > > I realise that I was misinterpreting "Requeue=1" as meaning that the job > had been requeued once, rather than that requeueing is allowed. > > However, I still don't see why setting > > #SBATCH -N 5 > > results in > > NumNodes=5-5 > > for a job which is still pending. Shouldn't it still be possible to run > the job on more than 5 nodes, if the number of tasks can be satisfied > that way? > > Cheers, > > Loris > > > > > > On Fri, Oct 19, 2012 at 11:00 AM, Loris Bennett > <[email protected]> > > wrote: > > > > Hi, > > > > I have noticed that for a job currently pending which was started with > > > > #SBATCH -N5 > > #SBATCH -n12 > > > > 'scontrol show job' gives > > > > Requeue=1 Restarts=0 BatchFlag=1 ExitCode=0:0 > > ... > > NumNodes=5-5 NumCPUs=12 CPUs/Task=1 ReqS:C:T=*:*:* > > > > Is the change of the number of nodes from '5' to '5-5' a result of the > > requeueing and, if so, is this the desired behaviour? > > > > I'm using version 2.2.7. > > > > Cheers, > > > > Loris > > > > -- > > Dr. Loris Bennett (Mr.) > > ZEDAT, Freie Universität Berlin Email > [email protected] > > > > > > -- > Dr. Loris Bennett (Mr.) > ZEDAT, Freie Universität Berlin Email > [email protected] -- Dr. Loris Bennett (Mr.) ZEDAT, Freie Universität Berlin Email [email protected]
