nvm

You have a rebase issue.
Don't use real urls in tests, mocking the time is much better as it doesn't depend on some server status. E.g. if chronik1.alitayin.com is down your second test is identical to the first one, and there is no way to tell

The title is still not good

Also please keep the title/summary up to date

A nit and a question, but it's getting there. Also you lost the test at some point, please bring it back.

It's adding code, not only testing it, please update the title/summary.

Please update the summary. The title is correct as it describes what is being done, but the summary does the same and should instead explain the rationale for this change. With no context the reviewer cannot decide if this is a good change or not. In this case you have 2 benefits:

• It simplifies the code because this function never used the FailoverProxy, and it shouldn't have been a member from the beginning.
• It makes it possible to reuse the function outside of the scope of the FailoverProxy, namely for the ping feature.

Please run the extended test suite. If done, update the test plan accordingly

In D17916#408501, @bytesofman wrote:

But for ECDSA signatures, the tests fail, as "real" signatures can have a different length than dummy signatures used for leftover calculation.

Does this mean we can't accurately estimate tx size of ECDSA -signed txs with dummy signatures?

Check against undesired txs being polled

Add children first to optimize the polls

@bot build-chronik

Rebase

Feedback

Rebase

Apart from a few nits and bad tests this code looks good.

As discussed offline the approach here is wrong. This code is trying to do load balancing, but this is a bad idea and should be handled server side and not client side: the server knows its load but the client can't.

Did you check for regression against an existing datadir ?