[Berzelius-users] Maintenance downtime 2024-04-03 -- 2024-04-04

[Henrik Henriksson]

Dear Berzelius users,

We are currently planning for a maintenance window of two full working days,
2024-04-03 07:00 -- 2024-04-04 18:00. This particular window was selected based
on the results of the downtime survey we conducted.

Inside of this window, the cluster will be completely unavailable. Please ensure
you copy out any data you need before the downtime starts. We may return the
cluster to operation earlier if possible, but don't count on this.

During the maintenance window, we will do the following:

   - Perform upgrades on our Lustre filesystem. This is the main reason for the
     downtime, as we need to bring the filesystem completely offline for this
     operation. We expect this to take two full days. From a user perspective,
     little will change, but this is an important infrastructure upgrade.

   - Update Slurm to a newer version. We don't expect any user-facing changes.
     The job queue should remain, unlike the previous maintenance window.

During the maintance window we may also try to fit in smaller tasks that require
less risk to do while the cluster is out of production anyway, such as firmware
updates. This will be minor changes that we don't expect to be user visible.

As always, please contact berzelius-support at nsc.liu.se with any questions or comments!

Kind regards,
Berzelius Staff