Apologies for the trouble. We had an incident affecting deployments across all regions, which has since been resolved. Your three build failures coincided with that incident, and your latest deployment is now succeeding. You can see the full timeline here: status.railway.com/incident/7M7CBL9X. Regarding the database, your Postgres service stayed up throughout, but its logs show lock and statement timeouts on a couple of your tables caused by query contention, not a crash. That is an application-level concern worth investigating in your query patterns and transaction handling.