Ceres cluster maintenance is scheduled for the week of February 20, 2023.
This will be a major upgrade, as Ceres is being upgraded from EL 7 to EL 9.
We will rebuild the software modules to run on the new OS. If you have installed software on your own, you may need to rebuild it under the new OS as well.
During the maintenance we will also upgrade File system which should resolve the issue with fasterq-dump locking up nodes.
Some network changes will be performed during the maintenance.
Users with no projects will be moved from scinet account to sandbox account. It’s expected that SCINet users have access to a project to perform meaningful computations on Ceres and/or Atlas. To request a project, go to our request resources page.
A lot of preparation work will be done prior to the maintenance, but we still may encounter problems during the update process.
We will send a follow up message when Ceres is available again.
Note that queued jobs will not start if they cannot complete by 6AM February 20. In the output of the squeue command the reason for those jobs will state (ReqNodeNotAvail, Reserved for maintenance). The jobs will start after the scheduled outage completes.
If you anticipate the need to access HPC resources during the week of February 20, 2023, we advise moving to Atlas cluster which will be available during the Ceres maintenance. Make sure to copy data from Ceres to Atlas prior to the maintenance if needed.
Please submit any questions you may have via email to firstname.lastname@example.org.