File server / central storage system: Maintenance announcement for Friday, 12.02.21 starting at 15:30

On the file server / central storage system, the RAM must be replaced on two storage nodes, as these exceed the guideline values for too many errors. The exchange is therefore to be carried out on Friday, 12.02.2021 from about 15:30. The exchange is expected to take 2 hours.

==================
Who is affected? (General)
==================

Potentially, all services / customers that have already established a connection to the two storage nodes are affected. New connections are not established to the two affected storage nodes and are therefore not affected. If you have had a connection open for some time, you can find out if you may be affected in the next section.

These services include: Home directories, shares / group drives and profiles (Windows), workgroup server, NEMO, bwLehrpool.

Service providers should therefore check whether the service is affected and take precautions to minimize the impact.

==================
Who is affected? (Technical)
==================

In order to keep the impact as low as possible, the two storage nodes have been taken out of the distribution system as a precautionary measure, so that no new connections are automatically distributed to these storage nodes (this is done by our load balancer).

However, existing connections will remain. This means that all services / customers that have a connection to the two storage nodes will potentially feel an impact. Therefore, as a precaution, please check if you are connected to the following IPs:

10.4.7.28
10.4.7.30
10.4.7.60
10.4.7.62
10.4.7.92
10.4.7.94
192.52.4.16
192.52.4.14

If you have a connection to these IPs, you will be affected by the maintenance.

Each storage node will be offline for at least 45 minutes, which means that your connection will potentially hang for 45 minutes (for NFS; for SMB the protocol should automatically reconnect to another storage node). Only when one storage node is back online, the second storage node will be repaired.

==================
Solution if you are affected
==================

If you have determined that you have a connection to the above IPs (or are unsure if you are affected), you can escape any degradation by establishing a new connection to the storage system. This will redirect you to an unaffected storage node through the load balancer.

SMB / CIFS (e.g. home directory as network drive) : This protocol actually connects automatically to a new storage node if the current storage node is no longer available. To avoid possible problems with an active data transfer you can proactively rebuild the connection before Friday to make sure you are not affected further.

NFS: If you have a hard mount, the connection will hang for the duration of about 45 minutes. You can proactively rebuild the connection before Friday to ensure you are not affected.

Please note that you should never connect directly to an IP address (e.g. 10.4.7.28). Always use the corresponding server address, e.g. files.uni-freiburg.de, ufr.isi1.public.ads.uni-freiburg.de , etc.

We apologize for any inconvenience this may cause and will do our best to keep the disruption to a minimum.

Best regards,
Your Storage Team