Closed ryan-summers closed 2 years ago
Yeah. We should probably not just stall incoming messages but actually discard reject them. Or mark them (republication messages) with a property as application-originating and discard based on that?
We could theoretically append some CorrelationData
indicating that they are republished messages.
While we can modify the Miniconf state machine to just reject/ignore any inbound settings configurations while doing republishing, this doesn't solve the problem completely since we would transition state after republishing the last setting, which then causes us to re-receive it. I think correlation data is likely the best approach.
When miniconf republishes settings, it registers the message that it has published as inbound change requests. This causes it to generate settings update events for all settings.