Sentry experienced a major outage due to Postgres' transaction ID (XID) wraparound issue, which can cause database corruption and data loss if not addressed promptly. The problem arises from the way Postgres handles XIDs, which are used to identify transactions, and the need for routine maintenance tasks called autovacuum to prevent this issue. Sentry's write-heavy application and large relational tables made it vulnerable to this problem, leading to a prolonged outage. To recover, Sentry had to shut down the database, truncate a critical table, and make aggressive hardware and configuration tuning changes to improve vacuum times. The experience highlights the importance of understanding Postgres' internals, tuning autovacuum effectively, and having robust safeguards in place to mitigate the risk of XID wraparound.