You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
fix: correct sled corruption recovery to remove actual db files
The previous recovery code tried to remove data_dir/state/ which doesn't
exist - sled stores files directly in data_dir (db, conf, snap.*, blobs/).
This caused validators with corrupted storage to restart in a loop
instead of auto-recovering.
Changes:
- Storage recovery now correctly removes sled files (db, conf, snap.*, blobs/)
- Preserves distributed-db directory (handled separately)
- ChainState recovery simplified - corrupted data overwritten on save
0 commit comments