Inactive replication slots prevent autovacuum from removing dead tuples cluster-wide

Inactive logical replication slots cause catalog_xmin to prevent autovacuum from cleaning dead tuples across the entire cluster. WAL files accumulate indefinitely, disk usage grows by hundreds of GB, and CPU usage remains high (~80% during business hours, ~35% off-hours) due to ineffective autovacuum repeatedly scanning tables without removing dead tuples.

Run 'SELECT * FROM pg_replication_slots;' to identify inactive replication slots. Drop unused slots with 'SELECT pg_drop_replication_slot(slot_name);'. Monitor disk usage immediately after removal - it should drop significantly as WAL files are purged. Verify autovacuum effectiveness by checking n_dead_tup in pg_stat_user_tables decreases over time. CPU usage should drop to <10% during business hours.