Reports have been made where Nexus has become non-responsive. Thread dumps suggest that in some cases threads were blocked or backing up waiting for connections from the Orient DB connection pool. This is not at the point where database queries are timing out - the diagnosis point is getting connections from the pool to make the query. Such conditions can be created if a poorly written Groovy script is not cleaning up after itself after throwing an exception and not closing transactions.
In order to make diagnosing the database connection pool state easier, adding an MBean that gets dumped into the jmx.json that captures the current state of the connection pool our database queries use would be a good idea and help us prove or disprove connection pool current state and provide guidance for possible adjustment of the pool.