Advice: GPFS server hazard: 2018-03-05 11:00 to 16:00

Follow

On Monday March 5th, QRIScloud and UQ staff will be applying patches to the GPFS servers that comprise the Medici data fabric.  These patches include fixes to problems that are leading to applications getting I/O errors when the fabric is under heavy load.  The work is due to start at 11am.  Patches will be applied one system at a time to try to minimize disruption.

We do not anticipate that there will be significant disruptions.  However:

  • Occasional service pauses of a few minutes are to be expected as GPFS related services cut over between HA server pairs following an upgrade.
  • If something goes wrong during the upgrade procedure, longer interruptions may occur.

The following systems may be impacted at different times:

  • The HPC systems: Awoonga, Flashlite and Tinaroo
  • The QRIScloud Nextcloud service
  • The "ssh" and "gridftp" access services (for RDS collections on GPFS)
  • Access to the Polaris GPFS home via servers at UQ St Lucia.
  • NFS access to collections on GPFS from NeCTAR VMs
  • Swift (object storage) access to collections on GPFS.
  • The QRIScloud Manilla NFS service.

We apologize for any inconvenience.  However, this is essential work, and it needs to be done during business hours.

UPDATE - 2018-03-05 11:00 - Upgrades commencing.

UPDATE - 2018-03-05 14:10 - Upgrades completed.

 

Have more questions? Submit a request

Comments

Powered by Zendesk