Make sleeve/fastdp status visible in Prometheus

[mikebryant]

Feature Request

Our Problem

We had a firewall rule in our test environment that allowed ipsec. This wasn't present in production. As a result weave downgraded all the connections to sleeve, leading to severe latency issues.

The request

Would it be possible to expose the sleeve/fastdp state in Prometheus? We'd like to add alerting so we can't be bitten by this situation in future without being told about it.

[bboreham]

I think you should be able to monitor for weave_flows going to zero, or maybe absent(weave_flows)

[bboreham]

But directly exposing the state is a great idea too.

[mikebryant]

Ah, thanks for the tip. Looks like that would be sufficient right now to give us an idea of status :)