Avatar
Peter Todd 2y ago

Good example of how stupid current lightning implementations are re: fees.

This tx paid $38/89,907sats to spend a single anchor output at 497 sats/vbyte (6x higher than necessary to get the tx mined), producing a single output just 6437 sats in size.

https://mempool.space/tx/158b0c7bfcdc32c4e3fed95c26b5b0f54fde5b7182d933fcff1563375e8ec77f

It's not a CPFP: the commitment transaction was mined 3 days before the anchor output was spent.

It's just an idiotic implementation that wasn't willing to forget about the anchor output; I know LND does this. The channel itself only had $2540 in capacity, so that one transaction spent 1.5% of the entire capacity on fees.

Reply to this note

Please Login to reply.

Discussion

Avatar
Michael Henke 2y ago

Would c-lightning handle this better? Alternative?

Avatar
Peter Todd 2y ago

Not sure. I don't run any CLN nodes myself. But I really should.

Avatar
Korhag 2y ago

The main problem with lightning is that I can’t understand it. Maybe even lightning won’t be the retail level stuff and it will be service. Medium sized players can make settlements with that but it might be still too complicated for coffee stuff.

Upward and onwards to Layer3! :)

Avatar
Carman 2y ago

By default it gets fees from bitcoind which always has terrible estimates

Avatar
Christopher 2y ago

Yes. Lightning implementations were built in a low fee environment and it shows. All of them overpay basically all the time, but the avoidable force closes due to implementation miscommunications are an even bigger issue.

But there is progress - and: The focus only shifts to these issues if they become real pain points.