September maintenance window closed
The September maintenance window will begin at 09:00 CEST on September 7.
- Queues on Rackham and Snowy will be stopped as we perform work on Crex.
- Queues on Miarka will not be stopped.
- The UPPMAX Cloud will be brought OFFLINE as we install GPUs.
- All systems will receive system updates and security fixes.
UPPMAX together with DDN will perform the last steps in completing the Lustre filesystem expansion of our storage system Crex, used by Rackham and Snowy for project and dataset storage. We will resize the metadata storage to allow for more inodes (= more files). The plan according to DDN is expected to be kept within a working day however final testing and assurance might push production onto Thursday, September 8.
The UPPMAX Cloud will (finally!) be equipped with 24 NVIDIA GPUs (20xA2 and 4xT4). The GPUs will bring much needed inference and light training capabilities to the UPPMAX Cloud. We will power off the compute hosts when installing the GPUs. If your VM or application requires special care before shutdown YOU must do so before 09:00 CEST on September 7.
Update 2022-09-07 09:00
Maintenance begins.
Update 2022-09-07 12:00
Maintenance is progressing as planned. File system metadata is currently being resized. The new GPUs has been installed.
Update 2022-09-07 15:00
We are working with the Crex metadata resize. Uppmax Cloud is up and running in production again.
Update 2022-09-07 17:00
Crex is up and running. We are working on the clusters.
Update 2022-09-07 09:00
Rackham is back in production. We are working on Snowy.
Update 2022-09-07 11:00
Snowy is back in production.