We are using Camunda 7.15 with an Azure Postgres DB. Occasionally Camunda loses its connection to Postgres. We have solved this by adding all the appropriate Camunda configurations:
ValidationQuery
ValidationQueryTimeout
TestWhileIdle
TestOnBorrow
RemoveAbandoned
This has reduced the incidents of connection closures and has also made Camunda more resilient to DB connection failures so the Camunda engine does not fall over if a connection is lost.
The problem we have now is that when we have intermittent connection closures external service tasks become stuck and any new service tasks are created but also become stuck. The only way to resolve this is to restart our c# backend which calls the service task rest API. This has led us to believe that the issue must be that somehow the polling of external tasks becomes deactivated when Camunda recovers fronm the Postgres connection closure.
This is causing us immense pain at the moment. Any pointers to it’s resolution would be gratfully received.