[Bi-users] Bi available again

[Kent Engström]

Kent Engström <kent at nsc.liu.se> writes:
> Dear Bi Users,
>
> since ~16:40 today we have issues with the InfiniBand network on Bi that
> means that most things do not work. We are troubleshooting this and will
> get back to you when we know more.

Short summary: Bi is now available again. All running jobs were lost and
you got kicked out from the login nodes while we where handling this.

Background: One of the 12 core switches in the Bi InfiniBand network
died and that was not handled gracefully. After attempts at less
intrusive interventions, we had to turn off all the core switches and
leaf switches to restart them again in an orderly manner. We have now
done that, restarted the nodes, checked that things seem OK, and opened
the system for user access again.

-- 
Kent Engström, National Supercomputer Centre
kent at nsc.liu.se, +46 13 28 4444