This message was deleted.

(I don't work here anymore and don't know where it might be documented publicly, probably nowhere) `management`: This is the 'original' type for all the clusters that are registered with rancher. Includes provisioning for everything except RKE2. `provisioning`: Newer, for a separate newer provisioning system. The

management

and

provisioning

types are synced both ways so that every cluster should have an entry in both, with the meat of the cluster spec in one and a stub in the other. This is specifically for only RKE2 in 2.6, other providers were supposed to be migrated over eventually but dunno if/when that's ever happening. `fleet`: Keeps track of the clusters tied to fleet ("Continuous Delivery"). Fleet is (at least in theory) a standalone project so it has its own list, but will generally contain the same clusters as above with fleet-specific status. `x-k8s.io`: The regular community k8s Cluster API. There's an RKE2 provider hooked up and

provisioning

and its controllers ends up creating these (and a bunch of related types), which ultimately creates the actual cluster resources in $your_infra_provider. `clusters.rancher.cattle.io`: Don't remember, and doesn't seem to exist in 2.6. Also weird, we usually avoided calling anything 'rancher' in the code/types… haven't upgraded my house to 2.7 yet.

Hello @full-painter-23916! Thank you for your answer! Indeed I have observed that the number of

clusters.provisioning

is identical with the number of

clusters.management

. I have also observed that in the majority of the cases the name of the two resources is identical. On the other hand I have also observed that there are some

cluster.fleet

resources that don't have a pair of the other two types. In what case could these appear?

Fleet is its own standalone project so you can register clusters with it that regular rancher control plane doesn't know about

Oh! I thought that they are more tightly integrated in these newer versions.