This message was deleted.

The first thing I'd try doing is playing with soft links so the two directory paths point at the same location. If the paths are on different mount points, at some point you probably need to shut down each node and actually move the data at the directory over to the new mount point, but in theory the other nodes should update it after you start back up and it'd be fine. If you're doing this because you're running out of space on

/var/lib

then you'd need to do the shut down & move immediately and make

/var/lib/etcd

the soft link to the other location.

Stop all etcd nodes. Change the data path. Use --cluste-reset to start the etcd on one node, and then restart the remaining nodes to rejoin the cluster? https://docs.k3s.io/cli/etcd-snapshot