In a scenario of a network outage or connection issue between Bitbucket and the Database for example, because the database connection pools are acquired within Bitbucket's startup, we are required to restart all of our nodes to restore the application.

Bitbucket as a enterprise application should have a self-healing mechanism to reconnect with the Database.

This would also be good to consider self heal for other components, like External Search, NFS and so on.

Instead of using db.pool.timeout.idle and configuring it from within the Database, it would be great if Bitbucket could self heal and reconnect its nodes with the database.