FATAL: could not write to file "pg_wal/..." — disk full (no Debezium error, PostgreSQL-side symptom)

Root cause

A lagging or disconnected logical consumer can retain WAL. Growth depends on actual database write volume and slot progress.

How to fix

1. Monitor pg_replication_slots, retained WAL bytes, database write rate and free disk.
2. Choose max_slot_wal_keep_size for the installed PostgreSQL version and recovery budget. Exceeding the cap can make the slot unusable; it is not a guarantee against disk exhaustion.
3. Review heartbeat.interval.ms and heartbeat.action.query with capture/publication filters; a heartbeat alone does not guarantee progress for every database workload.
4. Set alerts from remaining disk and expected recovery time, not a universal 1 GiB threshold.