There is a bug relating to calculating fees for next block templates. My assumption is that is extended to general fee estimates but I'm not sure. You cant rule out differences between bitcoin core fee estimates and non core estimates. Differences in them can cause force closures and we will see bigger differences with more active mempools like we have.
I notice some node runners, trying to increase their local mempool size in their umbrel/mynode/raspiblitz bundle node.
Could this affect also on fee estimates for their nodes?
reply
I don't think it would but I don't know the internals about how core estimates fees. It could be smart, or esplora could be smarter but if there's a big enough difference and not enough leeway then it would make lightning unusable across different estimation types.
reply