kartotherian is being moved to wikikube. When that is complete, the server backend will be moved to Bookworm.
We'll be reusing six former ganeti nodes a test cluster to test the new stack (setup at T380144) until we eventually reimage the maps* nodes.
The high level plan is:
- install a master bookworm node and fix all issues
- install a replica bookworm node and fix all issues
- the remaing replicas
- test the OSM import on the new cluster
- point wikikube/staging to use the bookworm cluster
- if this works fine, we can enable expiration events for bookworm and disable it for buster
- move the prod pods to the new cluster and eventually
- reimage the maps* from buster to bookworm
- failover production traffic to the actual maps* cluster