I experienced an unexpected Cloud SQL maintenance event outside the scheduled Maintenance Window. While logged into the GCP Console, I selected my SQL instance (PostgreSQL 16.9) and attempted to create a new user, but received an error stating that the operation was not allowed at the moment.
Although the database initially appeared accessible to my application, it stopped responding shortly after, resulting in application failures. About 10 minutes later, the instance status changed from “Maintenance” to “Runnable,” but it remained inaccessible. Under the Databases tab, the message displayed was: “Databases are currently unavailable.”
I manually restarted the instance, after which it returned to normal. The database was unavailable for approximately 25 minutes.
**Could you help me understand why this occurred and what might have caused it?
**
Instance details;
Region: europe-west3 (Frankfurt)
DB Version: PostgreSQL 16.9
It might be one of those rare cases where Cloud SQL has to perform urgent maintenance outside your scheduled window. This can happen when there’s a critical patch needed to fix stability issues or security vulnerabilities that can’t wait. While it’s not common, Google might push these updates quickly to protect your instance. Cloud SQL does count this type of downtime against the SLA, so it’s still tracked and acknowledged on Google’s end.
