We're planning to upgrade Slurm 17.02 to 17.11 soon, so it's important
for us to test the slurmdbd and database upgrade before doing the actual
upgrade.
I've made a *successful* upgrade of the database migration from 17.02 to
17.11, making a dry run on an offlined compute node running CentOS 7.4.
The dry run procedure has been documented in my Slurm Wiki page:
https://wiki.fysik.dtu.dk/niflheim/Slurm_installation#make-a-dry-run-database-upgrade
I tried Jessica's problem command, and it didn't cause any errors in my
case:
# sacctmgr -i delete account where account=XXX cluster=niflheim
At least it appears that we don't get hurt by Slurm Bug 4785 referred to
in the page mentioned below.
Question: Is it safer to wait for 17.11.4 where the issue will
presumably be solved?
On 02/14/2018 09:32 PM, Jessica Nettelblad wrote:
FYI - SchedMD has now solved the issue in the master branch.
https://github.com/SchedMD/slurm/commit/4a16541bf0e005e1984afd4201b97df482e269ee#diff-7649dde209b4e528e3ba8bb090b19f63
Best regards,
Jessica Nettelblad, UPPMAX
On Wed, Feb 14, 2018 at 3:31 PM, Bjørn-Helge Mevik
<b.h.me...@usit.uio.no <mailto:b.h.me...@usit.uio.no>> wrote:
Thanks for the heads-up! We're currently runnint 17.02.7 with MariaDB,
and haven't seen this problem, but we are going to upgrade to 17.11 in
the not-to-far-future.
/Ole
