FAS Research Computing - MPI - Critical security flaw must be patched, will break MPI cluster-wide – Incident details

System Under Maintenance

Status page for the Harvard FAS Research Computing cluster and other resources.

Cluster Utilization (VPN and FASRC login required): Cannon | FASSE


Please scroll down to see details on any Incidents or maintenance notices.
Monthly maintenance occurs on the first Monday of the month (except holidays).

GETTING HELP
https://docs.rc.fas.harvard.edu | https://portal.rc.fas.harvard.edu | Email: rchelp@rc.fas.harvard.edu

The colors shown in the bars below were chosen to increase visibility for color-blind visitors.
For higher contrast, switch to light mode at the bottom of this page if the background is dark and colors are muted.

MPI - Critical security flaw must be patched, will break MPI cluster-wide

Resolved
Operational
Started over 1 year agoLasted about 3 hours

Affected

Cannon Cluster

Operational from 3:55 PM to 7:02 PM

SLURM Scheduler - Cannon

Operational from 3:55 PM to 7:02 PM

Cannon Compute Cluster (Holyoke)

Operational from 3:55 PM to 7:02 PM

Boston Compute Nodes

Operational from 3:55 PM to 7:02 PM

GPU nodes (Holyoke)

Operational from 3:55 PM to 7:02 PM

FASSE Cluster

Operational from 3:55 PM to 7:02 PM

Updates
  • Resolved
    Resolved

    All the builds went pleasingly fast. MPI has been rebuilt and distributed. The clusters are back to normal.

    Thanks for your understanding in this and future matters where security is paramount.

    Thanks,
    FAS Research Computing
    https://www.rc.fas.harvard.edu

  • Update
    Update

    Security patching is complete (PMIx 5.0.1, Slurm 23.02.5). Partitions are re-opened and jobs are resumed.

    Rebuild of MPI across the cluster will take quite some time.

    ETA for MPI to be functional again is Thursday morning.

  • Identified
    Identified

    Patching has begun. Partitions are closed and all jobs suspended.

    ETA to return the cluster to operation is 1 hour.

    ETA for MPI to be functional again is Thursday morning.

  • Investigating
    Investigating

    The OpenPMIX project just released a set of fixes for a severe security issue. This will require an immediate update and about one hour of cluster outage. This flaw is rated as severe and is being treated as an emergency mission-critical flaw which must be addressed immediately.

    This will break MPI cluster-wide until we can fully rebuild all MPI stacks and Slurm integration. We will begin rebuilding immediately after patching, but this will take quite some time to build and deploy across the entire clusters (Cannon and FASSE).

    Current ETA is Thursday morning.

    We appreciate your understanding. Updates to follow here.

For issues not shown here, please contact FASRC via
https://portal.rc.fas.harvard.edu or email rchelp@rc.fas.harvard.edu