When a slot is inactive, that means that Murty and his team have stopped the replication on their side. The SQL below will identify inactive slots. We should schedule this script in a job , running every hour. (1) SELECT slot_name, pg_size_pretty(pg_wal_lsn_diff(pg_current_wal_lsn(),restart_lsn)) AS size, active, slot_type FROM pg_replication_slots; (2) Once you identify the replication slot, the job should drop it with the following command: SELECT pg_drop_replication_slot ('pg_replication_slot_name_from_above'); (3) Restart the database instance to clean up unwanted disk space. Notes from Google:- Thank you for contacting Google Cloud Support. My name is Jagadeesh and I will be assisting you with this issue. I understand that WAL logs are expanding to over 1.5TB and you want to know what is causing WAL logs to grow and you want to reduce it. Please confirm if ...