Ceph clusters will issue a health warning if the current running CRUSH tunables are not optimal for the current running Ceph version starting at v0.74.

In order to remove this warning from the Ceph cluster you can adjust the tunables on the existing cluster. Adjusting the CRUSH tunables will result in some data movement (possibly as much as 10% of the data on the cluster). This is the obviously preferred route to take, but care should be taken on a production cluster as any movement of data may affect the current cluster performance.
You can enable optimal tunables with:

ceph osd crush tunables optimal

If you begin to see issues with performance due to the load from the data movement on the cluster caused by the rebalance from the tunables change or if you run into a client compatibility issue (old kernel cephfs or rbd clients, or prebobtail librados clients) you can switch back to legacy tunables with:

ceph osd crush tunables legacy