WAL directory filling disk

Alert at 21:30: "DB host disk > 92% and climbing ~1%/min." The Postgres primary is on a 500GB volume. Writes are slowing and you're minutes from the disk filling, at which point Postgres will refuse writes and could crash. The disk dashboard shows the pg_wal directory growing fast, it's now 180GB, far above normal. Recent context: a replica went offline for maintenance 3 hours ago and hasn't reconnected, and someone set up a logical replication slot last week for a new analytics pipeline that's been failing silently. Table sizes themselves look normal. Triage and act fast.