Been there, it’s not fun. Transitioned the new HPC cluster from redhat to suse. Switched to Centos after a few years on suse due to pricing and their software becoming more and more unstable. The following cluster got CentOS from the start and then got migrated to rocky after they switched Centos to rolling release. It’s not easy running this stuff…
we’ve transitioned before from CentOS 7 to Rocky 8 and it was not too bad, mostly just a good amount of planning and testing. we’ve got about 160,000 cores between 2 clusters (1277 nodes that i help manage, and 825 that are located at a facility that handles hardware for us) – fortunately I do primarily hardware work so don’t deal with the software headaches myself, and the people that do are very capable - but typically you’ll have incompatibilities, and a lot of user support stuff for the first couple weeks until things smooth out.
Been there, it’s not fun. Transitioned the new HPC cluster from redhat to suse. Switched to Centos after a few years on suse due to pricing and their software becoming more and more unstable. The following cluster got CentOS from the start and then got migrated to rocky after they switched Centos to rolling release. It’s not easy running this stuff…
we’ve transitioned before from CentOS 7 to Rocky 8 and it was not too bad, mostly just a good amount of planning and testing. we’ve got about 160,000 cores between 2 clusters (1277 nodes that i help manage, and 825 that are located at a facility that handles hardware for us) – fortunately I do primarily hardware work so don’t deal with the software headaches myself, and the people that do are very capable - but typically you’ll have incompatibilities, and a lot of user support stuff for the first couple weeks until things smooth out.