D-oh. Chris and I sent the same query within a couple of minutes of one
another. Please ignore this one and respond to Chris Samuel's instead ("
Guidance on planning a slurmdbd outage").Regards Jeff Tan High Performance Computing Specialist IBM Research Collaboratory for Life Sciences, Melbourne, Australia Phone: +61 3 9035 4392 ----- Forwarded by Jeff Tan/Australia/IBM on 07/04/2014 16:28 ----- From: Jeff Tan/Australia/IBM To: "slurm-dev" <[email protected]>, Date: 07/04/2014 16:34 Subject: experience with temporary absence of slurmdbd on running system Hi! We're looking for some feedback on the length of time that slurmdbd can be absent on a running system, e.g., maintenance of the slurmdbd host. The documentation and some experience tells us that slurmctld will cache the accounting data in the meantime, but we're a bit cautious about how long this can be safely done for. We're looking at under an hour, but it's a centralized slurmdbd for four busy clusters. Any guidance out there? Thanks! Jeff Tan High Performance Computing Specialist IBM Research Collaboratory for Life Sciences, Melbourne, Australia Phone: +61 3 9035 4392
