Missing delegated routing client metrics

[2color]

Background

Someguy proxies provider requests to cid.contact by default.

However, we don't outgoing HTTP requests to the endpoint with any of the metrics. This means that if cid.contact is down, there's no way for us to know about this or alert.

Suggestion

Add a prometheus counter to track the number of requests and the status codes, along the lines of:

delegated_routing_client_request_count{code="200",host="cid.contact",route="/routing/v1/providers"} 100

[lidel]

That counter already exists, we should be able able to count the histogram events sum by (host) (rate(routing_http_client_latency_count{instance=~"$instance"}[$__rate_interval])) to see count per host.

This is similar bug to ipfs/rainbow#261 – Someguy seems to be missing /routing/v1 client metrics defined in boxo/routing/http/client/measures.go#L19-L20:

• routing_http_client_latency
• routing_http_client_length

We need them, to be able to visualize request/success rate per host.

[lidel]

We may be able to port fix from Rainbow: ipfs/rainbow#262

[lidel]

• Fixed in feat(metrics): add delegated routing client metrics to prometheus #113 / released in v0.9.1
• Metrics documented in https://github.com/ipfs/someguy/blob/main/docs/metrics.md