This issue has been resolved.

Background info update: During a standard, automated GCP server rotation some of our new servers were instantiated on IP ranges that weren't authorised to connect to the production databases. This has been resolved.

Our team continued to monitor the service and identified about 30% of connections were still receiving errors messages. This has now been fully resolved.

Our team has identified and resolved connection errors to the database, and is continuing to monitor the system

Investigating increased error rates connecting to database