Been swap free since 2005. Various systems and sizes, including the first sustainable 1PF system. Since the systems were diskless and stateless it made sense. Yes, systems can crash if out of RAM, but with stateless provisioning you just reboot. Personally I'd rather have a crash, rapid reboot stateless provision, and let another user get the node, than a job start to crawl and waste time on the system. Eventually the resource manager will kill it anyway for running out of time.
Anyway, that is my HPC baremetal answer. For virtualized I found that swap enabled me to pack more work on to fewer resources because I could safely overcommit the RAM. OTOH, that was not for HPC workloads. On Fri, Feb 6, 2015 at 3:35 PM, Prentice Bisbal <[email protected]> wrote: > Do any of you disable swap on your compute nodes? > > I brought this up in a presentation I gave last night on HPC system > administration, but realized I never actually did this, or know of anyone > who has. I would tweak the vm.overcommit_memory setting, but that's not the > same as disabling swap altogether. I'd like to try doing this in the future, > but I prefer to learn from someone else's mistakes first. > > > -- > Prentice > _______________________________________________ > Beowulf mailing list, [email protected] sponsored by Penguin Computing > To change your subscription (digest mode or unsubscribe) visit > http://www.beowulf.org/mailman/listinfo/beowulf _______________________________________________ Beowulf mailing list, [email protected] sponsored by Penguin Computing To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
