If a message fails to send, the plugin will write it to the db and try again later. However, this process is invisible unless you look directly in the db.

It would be nice to have the plugin export some api calls for queue management. E.g.

• a call to query the queue length and oldest msg (something like select count(id), min(created_ts) from proton_queue)
• a call that allows admins to manually trigger sending messages from the backlog (perhaps some kind of wrapper around handle_db_msgs). Without this, the system will only handle 100 at a time on calls that generate messages.