Generate 'snmp traps'

[kruisdraad]

Routers generate SNMP traps for diagnostics, now i am NOT saying we should use SNMP ... however we could use some kind of push notifications to send out diagnostics messages.

For example state changes as described in issue #456 would be a great candidate for this, to log this in some public / closed channel for ILP operators. You can see sessions flapping etc.

[adrianhopebailie]

There is currently a /metrics endpoint exposed when the admin-api is enabled. The output is formatted to be consumed by Prometheus (i.e. the output is polled)

A common pattern here would be to add a metric that is a state flag, probably one per plugin/account.
The only downside is this is a polling model as opposed to push and it could be that polling to frequently impacts the node if the other metrics are expensive to deliver.