Status:
10.08.26 All nodes are up. The faulty FibreChannel switch has been replaced.
27.07.2026, 14:00 stjerne3, stjerne5, stjerne14, stjerne15, nashira and vega are fixed. Still no solution to the FibreChannel switch issue, a second replacement switch is underway.
25.07.26, 16:30 Workstations and compute nodes are now open for login.
All linux machines now run the latest Red Hat 9.8 version and StorNext is running at version 7.2.5.
Remaining issues:
- We still need to replace a failed FibreChannel switch. This means that bandwidth to disk is reduced and tape backup is not working at the moment. We expect this to be fixed early next week.
- Due to a bug in the NVIDIA Infiniband software, Infiniband is not working on the compute nodes. We are waiting for a new version of the software to be released by the end of this month where this is fixed.
- We have some issues on stjerne3, stjerne5, stjerne14, stjerne15, nashira, vega and barakish that will be fixed on Monday.
- We need to install a new version of CUDA on the GPU machines.
25.07.26, 14:40 Still no solution to the FibreChannel switch issue. Work on workstations and cluster nodes progessing.
24.07.26, 23:00 There is some trouble configuring the replacement switch. Hitachi and Broadcom are working on it. We have had progress on updates of workstations and compute nodes, but work remains for Saturday.
24.07.26, 11:45 We have a broken FibreChannel switch that needs replacement. Hitachi will do that in a couple of hours when they get the replacement switch.
23.07.26, 18:50 New file systems created. Upgrade of nodes and workstations is for tomorrow
23.07.26, 17:00 Update of metadata controllers done. Now installing new storage
23.07.26, 09:45 Update of metadata controllers in progress
22.07.26, 19:02 Shutdown complete. Starting backup of metadata.
22.07.26, 18.02 Starting shutdown
22.07.26, 17:00 Shutdown in one hour
Downtime:
We are planning downtime on the computer system from Wednesday July 22. at 18.00 to Saturday July 25. 2026.
During the downtime we will:
- Integrate 1 PB new disk storage in the StorNext system
- Upgrade StorNext to version 7.2.5
- Upgrade Red Hat Linux to version 9.8 on compute nodes and workstations
- Upgrade NVIDIA Infiniband software on cluster nodes
- Upgrade the Infiniband swithces in the clusters with new firmware and software
- Upgrade network switches
During the downtime all compute nodes and workstations will be down, all StorNext storage will be unavailable and login to the system will be disabled.