Postgres Down or Slow: Disk Filled by WAL (pg_wal) — Space Recovery Order

What you'll see

• ERROR: could not write to file "pg_wal/...": No space left on device
• Checkpoints keep failing; the server pauses writes
• df shows the data volume at 100%, du pins it on pg_wal/

Root causes

Fix it

1. Free space safely — do NOT delete pg_wal files by hand

   # safe space: drop unused indexes? no — first: identify slots/archiver, then VACUUM isn't it either: move/drop other files or grow the volume; pg_wal files are needed for crash recovery

2. Drop the stale replication slot (the usual WAL hog)

   SELECT * FROM pg_drop_replication_slot('<slot>');   # only if the consumer is truly gone

3. Fix or disable the failing archive_command

   ALTER SYSTEM SET archive_command = '/bin/true';  -- temporary unblock; SELECT pg_reload_conf();

4. Force a checkpoint to recycle completed WAL segments

   CHECKPOINT;   # then watch: SELECT pg_size_pretty(pg_wal_lsn_diff(pg_current_wal_lsn(), '0/0')) AS lsn_age; df -h

5. Prevent recurrence: alerting + slot guards

   # monitor pg_replication_slots lag and disk at 80%; wal_keep_size sized to real retention needs