OTF Mattermost DB Store Errors

## Description Mattermost on Railway (project: otf-n8n-mattermost-umami) entered an 8-hour error loop on 2026-03-14 from ~07:14 to ~15:25 UTC. Every 1-2 minutes, the following errors repeated: - `Error occurred getting all pending statuses` - `Failed to fetch dnd statues from store` - `Failed to schedule job` - `Failed to get system bot` - `Unable to cleanup command webhook store` - `Unable to cleanup token store` - `Invalid session` with HTTP hijacked connection errors The service remained online in Railway but was functionally degraded — background jobs, session management, and status tracking all failed. ## Steps to Reproduce Three services (n8n, Mattermost, Umami) share a single Postgres instance on Railway. Under load, Postgres connection limits are exhausted, causing cascading failures in Mattermost's store layer. ## Resolution Service self-recovered around 15:25 UTC on 2026-03-14. No manual intervention required. Logs returned to normal from 2026-03-15 onward. **Recommendation:** Monitor shared Postgres connection usage. Consider adding `?connection_limit=10` to each service's DATABASE_URL or splitting into separate Postgres instances if this recurs. ## Related - Tech asset: [[otf-mattermost]]