Currently we're recycling the database connection pool every second, apparently to work around problems when the database goes away during a transaction. This should be handled by handling the error and retrying the operation and not trying to circumvent it by essentially opening a new connection on every request.
Currently we're recycling the database connection pool every second, apparently to work around problems when the database goes away during a transaction. This should be handled by handling the error and retrying the operation and not trying to circumvent it by essentially opening a new connection on every request.