Paused CDC Connector Left Replication Slot Holding WAL, Filling Postgres Disk
A Postgres primary database shut down after its disk filled, because a paused CDC connector's replication slot had kept all write-ahead log since it was stopped nine days earlier. Replication slots record the oldest WAL position a consumer needs, and by default Postgres retains that WAL without limit, so it keeps growing while the connector is idle. A quieter variant can occur even when the connector runs, if captured tables rarely change and the confirmed position lags behind. The article recommends heartbeat settings, including a dedicated heartbeat table in the captured database, plus Postgres safeguards such as max_slot_wal_keep_size and idle slot timeouts.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.
Discussion (0)
Log in to join the discussion and vote.
Log in