Hi,
On my backlog for quite some time, there was need to migrate existing CBS/koji builders to new(er) hardware/hosts. When I moved items from backlog to the new place, I so imported this one : https://pagure.io/centos-infra/issue/52
This mail is just a "FYI" as nobody should have noticed any service disruption, as each kojid instance was migrated while no job was running, so disabled, migrated, enabled back in the pool and so on.
Hopefully you still *should* notice some speed improvements, including for build jobs (newer/faster nodes) and surely because local storage for these builders is SSD backed (instead of old SAS disks on 8y+ nodes).
Same rule applies for the distrepo tasks, as we now have multiple nodes in the "createrepo" channel, with access to the koji NFS share now over 10Gbit ethernet (while it' was limited to 1Gbit on older nodes). That was possible because of the previous storage migration that happened last week : https://lists.centos.org/pipermail/centos-devel/2020-August/055981.html
Should you encounter any issue, (but we're ran several --scratch builds) , feel free to reply to this email, drop on #centos-devel irc channel on irc.freenode.net, or even better now (really!), create a ticket on https://pagure.io/centos-infra/issues.
Happy building !