The cleanup script used a pinned gcr.io/etcd-development/etcd:v3.5.9
image. Kind v0.32 ships etcd 3.6, which writes a different on-disk
format; cleanup ran against 3.5.9 produced data the new cluster's
etcd 3.6 couldn't read, breaking cluster recreation.
Capture the etcd image ref from crictl inside the Kind node right
after cluster creation, persist to {backup_dir}/etcd-image.txt, and
read it back in _clean_etcd_keeping_certs. Always matches the
version that wrote the on-disk format.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
832ab66df6