Dear de.NBI Cloud users,
the security updates of all GPU nodes have been completed. *You can now
start your GPU instances again.*
*Thank you* to everyone who shut down their GPU instances in time. This
made the maintenance much easier for us.
Instances that were still running at the start of the maintenance were
*shut down gracefully* by us before the hypervisors were updated. They
were not hard powered off.
1. Starting your GPU instances
Please start your GPU instances as usual, e.g. via the dashboard or with:
openstack server start
In a few cases, instances showed problems when they were started again
after the update. *If your instance does not boot, the GPU is not
available (e.g. nvidia-smi fails), or you notice any other problems,
please contact us* and include the instance ID and the project name. We
will help you get it running again.
2. Non-GPU instances: migration still in progress
The update of the remaining hypervisors is still in progress. Non-GPU
instances will continue to be live-migrated between hypervisors over the
next days. They keep running without a restart. Apart from *temporarily
reduced performance* during a migration, you should not notice much.
If you do notice problems, please let us know.
3. Please update your instances now
Please remember that our maintenance covers only the cloud
infrastructure. *You are responsible for the software inside your
instances.* Now that your GPU instances are running again, please
install all pending updates (kernel, PHP, Node.js, Python/Conda
environments, containers) as described in our previous mail, and
reboot. The same applies to all non-GPU instances.
If you have questions, please contact us at
*cloud@computational.bio.uni-giessen.de*.
Thank you again for your understanding and cooperation.
Best regards,
Frank Foerster
on behalf of the de.NBI Cloud Giessen team
Bioinformatics Core Facility, Justus Liebig University Giessen