Add a metric to track number of RPC nodes requests

[vpavlin]

---

Is your feature request related to a problem? Please describe.
It would be good to understand how many RPC node requests a Codex node is doing - to know what kind of RPC node does one need (i.e. is a free tier RPC node enough based on the needed rate limit?)

Describe the solution you'd like
I'd like have a Prometheus metric tracking number of RPC node requests, so that we can see some averages over time

Describe alternatives you've considered
Not sure, feels likes a basic counter metric is the most straightforward way:)

Additional context
Add any other context or screenshots about the feature request here.

[gmega]

Sounds like a reasonable request to me. I'll add to our backlog so we discuss it.

[AuHau]

It is reasonable ask but won't be that easy to implement. If you are trying to understand how much RPC calls Codex node uses on avarage, I would say @veaceslavdoina give you some answer.

The reason why it won't be that easy is because we use abstraction library nim-ethers that abstract the interaction with the RPC provider. So technically we would need to add some hook system there that would expose the raw RPC calls.

Not sure if there is some other easy way here, but I can't think of any.

[vpavlin]

Hmm..that makes sense, sounds like the easiest way then is to add prometheus metrics to nim-ethers (or any other underlying library). I assumes from the persepective of performance analysis it could be also interesting to not only know number of calls, but also amount of data transferred etc. and that definitely belongs to that underlying library

[veaceslavdoina]

If you are trying to understand how much RPC calls Codex node uses on avarage, I would say @veaceslavdoina give you some answer.

We are measuring that on RPC (Geth) side and for 10 Codex nodes we have the following right now

[vpavlin]

So 300k req/day per node? So in theory a free tier on Quicknode (https://www.quicknode.com/pricing) offering 10M req/month should still work...cool:)

[vpavlin]

Not sure if there is some other easy way here, but I can't think of any.

What if someone added Prom metrics to https://github.com/status-im/nim-json-rpc for every other project to benefit? (Waku also uses RPC to sync RLN tree...)

[AuHau]

Not sure if there is some other easy way here, but I can't think of any.

What if someone added Prom metrics to status-im/nim-json-rpc for every other project to benefit? (Waku also uses RPC to sync RLN tree...)

I doubt it being a library, they would like to have metrics (eq., HTTP server) as part of it. But for sure, there could be some supporting work done there, where there could be a callback that would be used every time there is an RPC call it would get a call and then it would be easy in the consuming apps to implement such a metric.