Active Router Health endpoint

[sanchaymittal]

Problem:

• Need transfer size based on active routers.

Example case/Test case:

• Nomad or xpolli can use this endpoint to figure out max transfer size based on active routers.

Owner:
@sanchaymittal @base0010

Proposal:

T0 [sdk]

• getActiveRouterHealth(): Call getRouterStatus() get live routers and call subgraphs to getRouterLiq().
• `getHealthFromEndpoint(): Call endpoint to get router health with available liq.

T1 [sdk-server]

• getRouterHealth(): Calls getActiveRouterHealth() to get router health and liq.
• expose this endpoint.

T0 [EC2][Cloudflare]

• host the server on ec2
• connect ec2 with Cloudflare

Test Cases:

- should work with endpoint
- should work if no endpoint
- should work if endpoint errors

[base0010]

[CF]:

• what kind of auth/endpoint does this need on the ec2 side? or none ? ie .is the ec2 internal only and exposed by CF to handle loadbalancing et al.
• CF endpoints ie. api spec:

• /
  • returns all router liveliness & liquidity indexed on address
• /?routerAddress=0x...abcdef
  • returns only that router's liveliness & liquidity

[sanchaymittal]

• Doesn't really need any auth for the ec2 endpoint, it's fine for anyone to access it.

[LayneHaber]

Is the reason we aren't using the subgraph for this because of the liquidity caching, the latency, or both? What would the body of the endpoint look like?

[sanchaymittal]

Yupp that too

[base0010]

** So I think we decided to just use AWS native loadbalancing etc. since this already will exist as an EC2.

** I do agree that there is mitigable concern with the router self reporting it's liquidity. Mitigation is to double check subgraph or just check subgraph instead.

** It would be ideal if you could have them sign this response in a way that convicts/slashed them for reporting incorrect liquidity. But that very much feels like a v2 idea.

[LayneHaber]

Staking/slashing for misreporting liquidity is probably not enforceable since its all offchain, I think if we need this to be for other things outside of the sdk (and it will grow to become the subgraph collation layer), then we should probably just use the subgraph but acknowledge it may not be in line with what the routers are saying.

I guess whether or not that's acceptable really depends on the other use cases (i.e. what is Nomad planning on using this for? For coinhippo this is likely fine.)

[jakekidd]

Staking/slashing for misreporting liquidity is probably not enforceable since its all offchain

It is if they sign a warrant when reporting it. Warrant being something that (we know) can be used to slash if the thing turns out to not be true

But also personally, I don't feel it's necessary. Routers can't bid if they don't have the money, and routers would want to report that they have as much as they do. Don't really see the incentive to lie?

EDIT: Actually maybe there's an incentive to lie and say they have a ton more just to get more bids (esp. cause multipath as well now).

[jakekidd]

We could aggregate the router API response and compare it actively to cached subgraph queries on the backend for our endpoint?