A scheduled database schema update triggered a metadata lock on our read-replica database, which prevented all tenant identification queries from completing. This caused the white label platform dashboards to become unresponsive, returning timeout errors for all dashboard login and UI requests.

Calls, automations, and API services were not affected. Only white label tenant dashboard access was impacted.

Root cause
A column addition migration executed successfully on the primary database but stalled on the read replica due to a pre-existing long-running query (1+ hours) that held a conflicting metadata lock. This created a cascading deadlock: the pending schema change blocked all subsequent read queries to the affected table, which is queried on every request for tenant resolution.

The read replica was restarted to clear the deadlocked state. The schema migration then applied cleanly, and all services resumed normal operation.