is etcd able to cache cluster or node info? delete var/lib/etcd and reinit stucks

[Kevinkevin189]

i init a cluster with three node using yaml config file, all conf is normal from the official guidance site.
then i deleted the etcd work folder, /var/lib/etcd, and use same config file, i decice to re-init a cluster with exactly same yaml conf file.
it failed, with so many errors, like "cluster id mismatch", and others,
is there any other thing that i didnt delete? or the etcd just memorized the config ,so i must change one?