Now Hawk is in full production, it is important to have a routine maintenance window to undertake core updates and modifications to the computing infrastructure. This will be a regular monthly maintenance window scheduled for the first Wednesday of each month.
The initial session is planned for Wednesday 6th February, during which time the cluster should be considered as running at risk, and might be unavailable during this maintenance period.
We would, therefore, advise that you do not expect the system to be available during this work. Dispatch of queued jobs will be resumed once the system is fully available
We will minimise the level of interruption to queued job execution by setting up a scheduler reservation for the duration of the work. Thus, jobs which cannot be finished in time will be kept in the queue and launched after the maintenance period rather than being launched & then interrupted before. This also means that there is likely to be capacity available for shorter jobs as we approach the maintenance period.
Given the notice period we hope that users will be able to work around any inconvenience that this may cause, but should you have any immediate concerns please contact us.
Please note, not all maintenance sessions will be necessary but they are important to allow us to schedule work to ensure the security and on-going performance of the Hawk cluster and minimise unscheduled service outages.
We will advise on equivalent work periods for Sunbird in the near future.