FAS Research Computing - Istorija obaveštenja

Svi sistemi funkcionišu

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
Documentation: https://docs.rc.fas.harvard.edu | Account Portal https://portal.rc.fas.harvard.edu
Email: rchelp@rc.fas.harvard.edu | Support Hours

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.
Funkcioniše

SLURM Scheduler - Cannon - Funkcioniše

Cannon Compute Cluster (Holyoke) - Funkcioniše

Boston Compute Nodes - Funkcioniše

GPU nodes (Holyoke) - Funkcioniše

seas_compute - Funkcioniše

Funkcioniše

SLURM Scheduler - FASSE - Funkcioniše

FASSE Compute Cluster (Holyoke) - Funkcioniše

Funkcioniše

Kempner Cluster CPU - Funkcioniše

Kempner Cluster GPU - Funkcioniše

Funkcioniše

FASSE login nodes - Funkcioniše

Funkcioniše

Cannon Open OnDemand - Funkcioniše

FASSE Open OnDemand - Funkcioniše

Funkcioniše

Netscratch (Global Scratch) - Funkcioniše

Home Directory Storage - Boston - Funkcioniše

Tape - (Tier 3) - Funkcioniše

Holylabs - Funkcioniše

Isilon Storage Holyoke (Tier 1) - Funkcioniše

Holystore01 (Tier 0) - Funkcioniše

HolyLFS04 (Tier 0) - Funkcioniše

HolyLFS05 (Tier 0) - Funkcioniše

HolyLFS06 (Tier 0) - Funkcioniše

Holyoke Tier 2 NFS (new) - Funkcioniše

Holyoke Specialty Storage - Funkcioniše

holECS - Funkcioniše

Isilon Storage Boston (Tier 1) - Funkcioniše

BosLFS02 (Tier 0) - Funkcioniše

Boston Tier 2 NFS (new) - Funkcioniše

CEPH Storage Boston (Tier 2) - Funkcioniše

Boston Specialty Storage - Funkcioniše

bosECS - Funkcioniše

Samba Cluster - Funkcioniše

Globus Data Transfer - Funkcioniše

Istorija obaveštenja

May 2025

Coldfront service disruption
  • Rešeno
    UTC
    Rešeno
    This incident has been resolved.
  • Istražuje se
    UTC
    Istražuje se

    Coldfront is undergoing maintenance and may experience service disruption in the course of this process

    ETA for resolution is end of business on Friday.

Starfish upgrade Thursday, May 29th from 5PM-6PM
  • Završeno
    May 30, 2025 u 2:27 PMUTC
    Završeno
    May 30, 2025 u 2:27 PMUTC
    Maintenance has completed successfully.
  • Planirano
    May 29, 2025 u 9:00 PMUTC
    Planirano
    May 29, 2025 u 9:00 PMUTC
    Starfish upgrade Thursday, May 29th from 5PM-6PM. Starfish will be unavailable during that time
Coldfront is undergoing necessary maintenance. WIP
  • Rešeno
    UTC
    Rešeno
    maintenance complete
  • Identifikovano
    UTC
    Identifikovano
    Coldfront is in need of database maintenance. Expect return to service by 5PM.
MGHPCC power work 5/21 - 5/23 - Some partitions will be at half capacity
  • Završeno
    May 23, 2025 u 7:00 PMUTC
    Završeno
    May 23, 2025 u 7:00 PMUTC
    Maintenance has completed successfully
  • U toku
    May 21, 2025 u 11:00 AMUTC
    U toku
    May 21, 2025 u 11:00 AMUTC
    Maintenance is now in progress
  • Planirano
    May 21, 2025 u 11:00 AMUTC
    Planirano
    May 21, 2025 u 11:00 AMUTC

    The MGHPCC Holyoke data center will be performing power work on May 21st -23rd. This work will take out one half (or one 'side') of the power capacity for certain rows/racks including our compute rows. Because of our power draw, one side is not enough power to keep each full rack running.

    As such, we will be adding a reservation to idle half the nodes in the partitions listed below. A reservation will cause nodes to drain as jobs complete and stop scheduling new jobs on those nodes if they cannot be completed before the outage. This will allow us to idle and power down those nodes prior to the work and avoid potential blackout/brownout on those racks.

    This will mean that these partitions will be up and available, but that half the nodes from each will be down (assuming an even number of nodes).

    This work is part of an on-going power capacity upgrade at MGHPCC. We expect this will be the last power work needed and the facility will then provide enough additional power for future expansion as well adding overhead for the current load.

    The affected partitions are:

    • arguelles_delgado

    • bigmem_intermediate

    • blackhole_gpu

    • eddy gershman

    • hejazi

    • hernquist

    • hoekstra

    • huce_ice

    • iaifi_gpu

    • iaifi_gpu_requeue

    • iaifi_priority

    • jshapiro

    • jshapiro_priority

    • kempner

    • kempner_requeue

    • kempner_h100

    • kempner_h100_priority

    • kempner_h100_priority2

    • kovac kozinsky

    • kozinsky_gpu

    • kozinsky_requeue

    • ortegahernandez_ice

    • rivas

    • seas_compute

    • seas_gpu

    • siag_combo

    • siag_gpu

    • sur

    • zhuang

Issue with Jupyter app for OOD
  • Rešeno
    UTC
    Rešeno

    We have reverted the changes to the Jupyter OOD app while we work on further development.

    This incident is resolved.

  • Identifikovano
    UTC
    Identifikovano
    The Jupyter app for OOD is failing to start for certain workflows that load older versions of python or modules. We are in the process of reverting Jupyter to fix the bug.

Apr 2025

Starfish login unavailable
  • Rešeno
    UTC
    Rešeno
    Starfish is once again accepting logins.
  • Istražuje se
    UTC
    Istražuje se

    Starfish is not allowing login due to a database issue. We have contacted the vendor and expect to resolve this soon.

Power Loss at MGHPCC
  • Rešeno
    UTC
    Rešeno

    Most nodes have been returned to service, both Cannon and FASSE are operable. Remaining down nodes will be remediated by FASRC staff over the next several days.

  • Istražuje se
    UTC
    Istražuje se

    At 1:05am EDT MGHPCC lost power to compute due to cooling failure. The chiller has been restored and power is back as of 5:43am. As a result of this all compute is powered off and all running jobs were requeued. FASRC staff is triaging the down hardware and will bring it back online when it is ready. Jobs will start once FASRC staff have confirmed that everything is in good state.

holylabs reboot to resolve instability
  • Rešeno
    UTC
    Rešeno

    holylabs is back up

  • Identifikovano
    UTC
    Identifikovano
    holylabs must be rebooted to resolved a stability/funciton issue.
Login nodes temporarily down
  • Rešeno
    UTC
    Rešeno

    Cannon boslogin and FASSE login nodes are back up and operational.

    All holylogin nodes are still down for repair, please see our posted incident for more updates: https://status.rc.fas.harvard.edu/cm97gyay90013dturk7fxg5pb

    We apologize for the unexpected disruption.

  • Istražuje se
    UTC
    Istražuje se

    Due to a configuration error, all cluster login nodes are rebooting and are temporarily unavailable. Please save any work immediately.

holylogin[05-08] down
  • Rešeno
    UTC
    Rešeno
    Hardware has been repaired and holyoke login nodes are back online. Thanks for your patience.
  • Nadgledanje
    UTC
    Nadgledanje

    Holylogin chassis repair during maintenance was unsuccessful and replacement parts have been ordered.

    • holyoke login nodes (holylogin05-08) are down for hardware repair

    • Only Boston login nodes available (ie, boslogin[05-08])

    If you have holylogin hard-coded in your scripts, please update to login.rc.fas.harvard.edu or boslogin.rc.fas.harvard.edu for the time being, which will redirect you to an available login node.

    As always, the best method for obtaining a login node is usinglogin.rc.fas.harvard.edu which will pick a node for you.

    If you require a login node in a specific data center, use boslogin.rc.fas.harvard.edu (Boston) or (once they are back in service) holylogin.rc.fas.harvard.edu (Holyoke).

    See also: Command line access with Terminal (login nodes) – FASRC DOCS

  • Rešeno
    UTC
    Rešeno

    This incident was posted by mistake.

    holylogin01-04 were replaced by holylogin05-08 some time back.

    As always, the best method for obtaining a login node is usinglogin.rc.fas.harvard.edu which will pick a node for you.

    Or if you require a login node in a specific data center, use boslogin.rc.fas.harvard.edu (Boston) or holylogin.rc.fas.harvard.edu (Holyoke).

    See also: Command line access with Terminal (login nodes) – FASRC DOCS

  • Istražuje se
    UTC
    Istražuje se

    Holylogin chassis repair during maintenance was unsuccessful and replacement parts have been ordered.

    Audience:

    • All cluster users

    Impact:

    • All holylogin** servers will be down till further notice

    • Only Boston login nodes available (ie, boslogin[05-08])

    If you have holylogin hard-coded in your scripts, please update to login.rc.fas.harvard.edu or boslogin.rc.fas.harvard.edu for the time being, which will redirect you to an available login node.

    Updates to follow as we have them.

Mar 2025

seas_gpu partition occasional pre-emption issue
  • Rešeno
    UTC
    Rešeno

    This issue has not reoccurred and appears to be due to unique cluster state. We are resolving this but will continue to monitor for recurrence.

  • Identifikovano
    UTC
    Identifikovano
    We are investigating an issue with the seas_gpu queue where backfill jobs sometimes are not pre-empted. We have filed a bug with SchedMD and are monitoring.
Starfish maintenance - Friday March 21, 2025 5pm-7pm
  • Završeno
    March 21, 2025 u 11:00 PMUTC
    Završeno
    March 21, 2025 u 11:00 PMUTC
    Maintenance has completed successfully
  • U toku
    March 21, 2025 u 9:00 PMUTC
    U toku
    March 21, 2025 u 9:00 PMUTC
    Maintenance is now in progress
  • Planirano
    March 21, 2025 u 9:00 PMUTC
    Planirano
    March 21, 2025 u 9:00 PMUTC

    Starfish will be under maintenance Friday March 21, 2025 5pm-7pm

    Starfish will be inaccessible during this period.

boslfs02 hanging issues - rebooting
  • Rešeno
    UTC
    Rešeno
    boslfs02 has been restarted
  • Identifikovano
    UTC
    Identifikovano
    boslfs02 is experiencing issues. We are restarting the filesystem.
Open OnDemand update March 12th 2025, 8:30AM-11:30AM
  • Završeno
    March 13, 2025 u 3:30 AMUTC
    Završeno
    March 13, 2025 u 3:30 AMUTC
    Maintenance has completed successfully
  • U toku
    March 13, 2025 u 12:30 AMUTC
    U toku
    March 13, 2025 u 12:30 AMUTC
    Maintenance is now in progress
  • Planirano
    March 13, 2025 u 12:30 AMUTC
    Planirano
    March 13, 2025 u 12:30 AMUTC

    OOD will be updated to include a Metrics page, reachable from the Help menu

    • Audience: Affects all production Open OnDemand nodes on FASSE and Cannon

    • Impact: There will be minimal impact to users. Some users may encounter an error on the dashboard. Simply refresh the page a few times and the error should go away.

    Logins and jobs on OOD should not be affected during this maintenance.

FASRC monthly maintenance - Monday March 3rd, 2025 from 9am-1pm
  • Obaveštenje
    March 03, 2025 u 6:00 PMUTC
    Obaveštenje
    March 03, 2025 u 6:00 PMUTC
    Maintenance has completed successfully
  • Završeno
    March 03, 2025 u 6:00 PMUTC
    Završeno
    March 03, 2025 u 6:00 PMUTC
    Maintenance has completed successfully
  • U toku
    March 03, 2025 u 2:00 PMUTC
    U toku
    March 03, 2025 u 2:00 PMUTC
    Maintenance is now in progress
  • Planirano
    March 03, 2025 u 2:00 PMUTC
    Planirano
    March 03, 2025 u 2:00 PMUTC

    PLEASE NOTE - New time window going forward - 9am-1pm

    FASRC monthly maintenance will take place Monday March 3rd, 2025 from 9am-1pm

    NOTICES

    • Training: Upcoming training from FASRC and other sources can be found on our Training Calendar. at https://www.rc.fas.harvard.edu/upcoming-training/

    • Status Page: You can subscribe to our status to receive notifications of maintenance, incidents, and their resolution at https://status.rc.fas.harvard.edu/ (click Get Updates for options).

    • Upcoming holidays: Memorial Day - Monday, May 26

    • You can subscribe to our status page using the Get Updates button in the upper right

    MAINTENANCE TASKS
    Cannon cluster will be paused during this maintenance?: YES
    FASSE cluster will be paused during this maintenance?: YES

    • Slurm Upgrade to 24.11.2 - Crucial Update

      • Audience: All cluster users

      • Impact: Jobs and the scheduler will be paused during this upgrade

    • Open Ondemand (OOD) reboots

      • Audience: All OOD users

      • Impact: All Open OnDemand (aka OOD/VDI/RCOOD) nodes will be rebooted

    • Login node reboots

      • Audience: Anyone logged into a FASRC Cannon or FASSE login node

      • Impact: Login nodes will rebooted during this maintenance window

    • bos-Isilon firmware updates

      • Audience: bos-isilon users

      • Impact: No noticeable impact for storage users

    • Netscratch retention/cleanup ( https://docs.rc.fas.harvard.edu/kb/policy-scratch/ )

      • Audience: Cluster users

      • Impact: Files older than 90 days will be removed. Please note that retention cleanup can and does run at any time, not just during the maintenance window.

    Thank you,
    FAS Research Computing
    https://www.rc.fas.harvard.edu/
    https://docs.rc.fas.harvard.edu/

Mar 2025 do May 2025

Sledeći

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