[Berzelius-users] Cluster available

[Henrik Henriksson]

Dear Berzelius users,

Berzelius is now available again, however, with some caveats.

Due to some issues that surfaced during the move to our new cluster management
stack, parts of the migration got delayed. However, most users should now be
able to work normally.

Currently known user-facing issues:

   - Accounting data is currently not sent to SUPR, use `projinfo` on the
     cluster. As it's a new slurm installation, we start from zero.

   - `jobgraph` is currently unavailable. `nvtop` is available on the nodes.

   - The planned MIG-reservation is currently unavailable.

   - The fat nodes may remain unavailable overnight.

We expect to resolve these known issues during the next few days.

As many things have changed, we also expect new issues to crop up. In principle,
this is a new cluster. Please report any bugs or issues you find to
berzelius-support at nsc.liu.se !

All in all, the migration has gone mostly as planned, and we are already finding
the new management software easier to work with. While users should not directly
notice much difference, these changes will (eventually) result in a better
service.

--
Henrik Henriksson
Systems Administrator
National Supercomputer Centre