QRIScloud users are advised that on July 31st and August 1st there will be a series of rolling outages affecting most QRIScloud NeCTAR instances. These outages are part of the NeCTAR-wide hypervisor upgrades announced here.
The operating systems on all QRIScloud NeCTAR compute node need to be upgraded to Ubuntu Xenial, and this requires downtime for the instances. The upgrade work will be performed progressively, with a small number of nodes out of service at any one time. The upgrade procedure for each node will be as follows:
- All ACTIVE (i.e. running) NeCTAR instances on the compute node will be shut down.
- The compute node operating system will be upgraded, and the compute node will be rebooted.
- All instances that were ACTIVE in step 1. will be restarted.
We have sent emails to the Tenant Managers and Tenant Members for all instances that will be affected by the outages. Any other related outages to QRIScloud services will be announced separately. QRIScloud HA instances should not be affected.
We cannot give a detailed schedule for when specific instances will be shut down and restarted, but we will use the @QRIScloud twitter feed and this announcement page to keep you posted on progress.
- If your instances are providing services to end-users, plan an outage window, and announce it to them.
- Ensure that your critical instances have appropriate (and tested) backup / disaster recovery procedures, and that you have recent backups prior to the outage window.
- We recommend that you ensure that your instances are in SHUTDOWN state before the outage window, and then restart them yourself when notified. You should definitely do this if your instances are interdependent, or if they have to be restarted in a specific order.
- If you have instances that are surplus to requirements, please release the resources for other users to use. This is includes instances that have been in SHUTDOWN, PAUSED, STOPPED or ERROR state for a long time. Please read QRIScloud FAQ 5.3.5 for information on how to safely Terminate an instance.
If you have concerns about these outages, please contact QRIScloud support.
UPDATE - upgrades completed.
All 56 regular NeCTAR QRIScloud compute nodes have been upgraded to Xenial.