How can you recover safely from a failed PostgreSQL pg_upgrade when using link mode?
After a pg_upgrade operation fails, the administrator may be left with an inaccessible old cluster when link mode was used, or a partially upgraded state that prevents starting the new server. The goal is to determine a safe recovery path that preserves data integrity without requiring a full dump/restore, while considering constraints such as filesystem lay
foundation