Tier-1 sofia changelog#

2026-11-06#

  • Roll-out of core system upgrades to all compute and login nodes. This update includes Linux kernel version 5.14.0-687.65.1.el9_8.x86_64, GPFS storage driver version 5.2.3.10, DOCA-OFED network driver version 3.5.0 and NVIDIA GPU driver version 620.70.06.

2026-10-28#

  • Redundancy tests carried out on power and network systems.

  • Upgraded the firmware of zen4_h200 compute nodes.

2026-10-27#

  • Enabled module browser in portal.sofia.vub.be

  • Installed software toolchain foss/2026.1

2026-10-21#

  • Roll-out of core system upgrades to all compute and login nodes. This update includes Linux kernel version 5.14.0-687.58.1.el9_8.0.1.x86_64, GPFS storage driver version 5.2.3.10, DOCA-OFED network driver version 3.5.0 and NVIDIA GPU driver version 615.71.10.

  • Upgraded Open OnDemand on portal.sofia.vub.be to version 4.2.5

  • Enabled support for GPUs on EESSI software stack.

2026-10-20#

  • Replaced broken memory DIMM on node hc011.

2026-10-13#

  • Started daily storage snapshots of user home directories under /user/sofia and project folders under /sofia/projects. Snapshots are held for 7 days.

  • Installed mysnapshots tool on the login and compute nodes to check and recover lost data.

2026-10-01#

  • Roll-out of core system upgrades to all compute and login nodes. This update includes Linux kernel version 5.14.0-687.42.1.el9_8, GPFS storage driver version 5.2.3.9, DOCA-OFED network driver version 3.5.0 and NVIDIA GPU driver version 610.57.04.

2026-09-29#

  • Waldur site agent updated with sofia-storage plugin version 0.5.0

2026-09-16#

  • IOMMU pass-through enabled on zen5-dense and zen5-himem nodes for improved MPI performance.

  • Added specific QoS for zen5-vis to limit user allocation of resources on the visualization nodes.

2026-09-07#

  • Slurm updated to version 25.11.8.

  • Enabled support for GDR Copy on all GPU nodes.

2026-08-12#

  • Insulation of the main cooling loop completed.

2026-08-10#

  • Motherboard replaced on node hm001.

2026-08-04#

  • Planned intervention on the cooling system: water leak on rack 3 fixed.

  • Memory module replaced on node hc034.

2026-07-24#

  • Updated Open OnDemand to version 4.2.3.

2026-07-21#

  • Updated the kernel to version 5.14.0-687.26.1.

  • EESSI made available to all users.

2026-07-15#

  • Planned intervention on cooling: firmware upgrade of the CDU and fix the CDU monitoring.

2026-07-13#

  • Created a new Globus collection for sofia: the existing one had an incorrect configuration that caused absolute paths to not work. The previous is marked as legacy and will be removed by October 1.

2026-07-09#

  • Planned intervention on the cooling system.

  • Compute nodes shut down for 6 hours (unscheduled) to fix a water leak in the cooling system.

2026-07-07#

  • sofia enters production.

2026-07-06#

  • Operating system upgraded to Rocky Linux 9.8.