Schema operations fail with 'class already exists' after collection recreation

After deleting and attempting to recreate a collection with the same name, TYPE_ADD_CLASS operations fail with 'class already exists' error. Indicates stale schema state persisting across deletion/recreation cycle, requiring pod restart to clear.

1) After collection deletion timeout, wait for schema propagation across all nodes before recreation (check logs for TYPE_DELETE_CLASS completion). 2) If 'class already exists' error occurs, verify collection is truly absent via schema endpoint. 3) Restart pods to clear stale schema cache. 4) Avoid immediate collection recreation with same name after deletion - use unique names or wait for full propagation. 5) Upgrade to v1.26.13+ for improved schema consistency.