Cross-database lock information incomplete in pg_locks

pg_locks provides a global view of locks across the entire cluster, but joining its relation column with pg_class.oid only works correctly for relations in the current database (where database column equals current database OID or zero for shared objects). Cross-database lock analysis will return incomplete or incorrect results.

When analyzing locks across multiple databases, filter pg_locks results by the database column to match the current database OID or use database-specific connections. Do not attempt to join relation information across databases. For cluster-wide lock analysis, query pg_locks from each database separately or focus only on shared objects (database=0).