MBS currently ingests messages from outside systems from the message bus, but messages MBS creates just to trigger certain handlers are directly put on its internal python queue. This becomes an issue when MBS is restarted in the middle of processing these messages, as the python queue gets emptied as part of the restart. When using a message bus that supports queues, this issue is solved since the messages would still be in the queue when MBS reconnects to the message bus.
Alternatively, we could implement the "internal" queue as a redis queue or some similar technology.
redis
Metadata Update from @mprahl: - Issue assigned to yashn
Login to comment on this ticket.