If the goal is to allow jobs to request individual CPUs instead of individual nodes, you'll likely want to use select/cons_res. http://slurm.schedmd.com/cpu_management.html and http://slurm.schedmd.com/cons_res.html
- Trey ============================= Trey Dockendorf Systems Analyst I Texas A&M University Academy for Advanced Telecommunications and Learning Technologies Phone: (979)458-2396 Email: [email protected] Jabber: [email protected] On Tue, Jun 9, 2015 at 12:11 PM, Zentz, Scott C. <[email protected]> wrote: > Ok, I tried both DefMemPerNode=6000 and DefMemPerCPU=6000 and neither > option allowed for more than one job per node. What plugin are most people > using for their “SelectType”? > > > > Thanks! > > -scz > > > > > > > > > > > > *From:* Will French [mailto:[email protected]] > *Sent:* Tuesday, June 09, 2015 11:41 AM > *To:* slurm-dev > *Subject:* [slurm-dev] Re: Set 1 job per core. > > > > > > On Jun 9, 2015, at 10:16 AM, Zentz, Scott C. <[email protected]> wrote: > > > > Yes, I have > > > > DefMemPerCPU=6 > > > > Should that be specified in Kb? > > > > No, that’s in MB. I see you’re using SelectType=select/linear. Others with > experience with this option would probably be more helpful, but you might > try configuring DefMemPerNode instead of DefMemPerCPU as the SLURM man > pages state: > > > > DefMemPerNode > Default real memory size available per > allocated node in MegaBytes. > Used to avoid over-subscribing memory and causing > paging. DefMem- > PerNode would generally be used if whole nodes are allocated > to jobs > (SelectType=select/linear) and resources are shared > (Shared=yes or > Shared=force). If not set, > the DefMemPerNode value for the entire > > cluster will be used. Also see DefMemPerCPU and MaxMemPerNode. > > DefMemPerCPU and DefMemPerNode are mutually exclusive. NOTE: > Enforcement of memory limits currently requires enabling of > account- > ing, which samples memory use on a periodic basis (data > need not be > stored, just collected). > > >
