Dear list, Apologies. It seems the nodes I *thought* I had updated to SLURM 21.08 were not yet updated when I deployed the new configurations. Ouch! Shortly after the cancelling and rescheduling mishap I updated the nodes properly and now they recognize the new AccountingStoreFlags=job_comment configuration option. Sorry for the confusion. A painful lesson to learn!
Regards, On Sun, Nov 28, 2021 at 2:32 PM Alan Orth <[email protected]> wrote: > Dear list, > > I just upgraded my cluster from SLURM 20.11.8 to 21.08.4. Before the > upgrade I updated my configuration based on this comment from the release > notes¹: > > > -- Removed AccountingStoreJobComment option. Please update your config > to use > > AccountingStoreFlags=job_comment instead. > > After updating the slurmd.conf I upgraded SLURM, but got this error: > > > slurmd[21264]: error: _parse_next_key: Parsing error at unrecognized > key: AccountingStoreFlags > > slurmd[21264]: error: Parse error in file /etc/slurm/slurm.conf line > 119: "AccountingStoreFlags=job_comment" > > slurmd[21264]: fatal: Unable to process configuration file > > Then slurmctld drained all my nodes and all my jobs got cancelled. After I > removed the invalid AccountingStoreFlags option and restarted the SLURM > daemons on all nodes the jobs got rescheduled, but now all nodes are > drained due to "Duplicate jobid". *sigh*. > > What happened here? Is this a bug? This is the messiest SLURM upgrade I've > had in years... thank you for any advice, > > ¹ https://github.com/SchedMD/slurm/blob/slurm-21.08/RELEASE_NOTES#L135 > > -- > Alan Orth > [email protected] > https://picturingjordan.com > https://englishbulgaria.net > https://mjanja.ch > -- Alan Orth [email protected] https://picturingjordan.com https://englishbulgaria.net https://mjanja.ch
