CHPC - Nvidia GPU driver critical update – Incident details

Nvidia GPU driver critical update

Resolved
Major outage
Started 5 months agoLasted 43 hours 5 minutes

Affected

General Environment (GE)

Partial outage from 11:00 PM to 6:05 PM

HPC clusters

Partial outage from 11:00 PM to 6:05 PM

Windows servers

Partial outage from 11:00 PM to 6:05 PM

Virtual machines (VMs)

Partial outage from 11:00 PM to 6:05 PM

Protected Environment (PE)

Partial outage from 11:00 PM to 6:05 PM

HPC clusters

Partial outage from 11:00 PM to 6:05 PM

Updates
  • Resolved
    UTC
    Resolved
    This incident has been resolved.
  • Update
    UTC
    Update

    All the HPC cluster GPU drivers were updated and queues were released. Most of standalone servers have been updated as well.

  • Update
    UTC
    Update
    We are continuing to work on a fix for this incident.
  • Identified
    UTC
    Identified

    Nvidia released new GPU drivers that address critical vulnerability, https://nvidia.custhelp.com/app/answers/detail/a_id/5821

    All systems with Nvidia GPUs is affected. We will be updating the drivers, or have updated them already, on HPC clusters, owner interactive nodes such as the Cryosparc machines, and Virtual Machines. The processes running on GPUs have to be killed for the driver update to apply. On the Linux systems, the update should not require a reboot, but, it's possible that some machines will have to be rebooted.

    If you have any questions, please, contact our helpdesk, helpdesk@chpc.utah.edu