36 sats \ 5 replies \ @petertodd 11 Feb 2023 \ on: Discussion about increasing default mempool size limit in Bitcoin Core bitcoin
I don't think this is a good idea.
There's no reason to have tons of transactions floating around that aren't going to be mined in a reasonable amount of time. Better that users broadcast transactions that make financial sense, and use up network bandwidth for those transactions only. That works best when the min fee rate actually matches what is going to be mined in a reasonable amount of time, and the min fee increases appropriately.
Also, the rational given re: transaction expiration doesn't make any sense. The length of time a transaction can stay around in the mempool isn't directly related to the size of the mempool at all; the mempool is not a first-in-first-out (FIFO) queue. It's a priority system where whomever pays the most, gets served the fastest. The idea that "300MB is 2 days of transactions" is completely wrong.
Of course, on an individual level, you can use whatever mempool size you want. You'll just waste a bit of bandwidth compared to using a size that your peers are using.
(When I get home I need to login to GitHub and add these comments to the pull-reqs...)
aren't going to be mined in a reasonable amount of time
Question is what is reasonable amount of time here. 300 MB with segwit is potentially less than 24h. Main impact IMO is that tx below the drop feerate for most of the nodes are simple to replace, without fee bumping.
reply
What are your calculations for "less than 24hr"?
You can't do that math by assuming transactions are processed first-in-first-out. That's not how the mempool works.
I created a transaction a few days ago at 1sat/vbyte myself. It's still waiting, because obviously, why wouldn't it? The fee market has been such that it's never been worth mining a 1sat/vbyte tx.
reply
I meant assuming no new transactions comes in, what could fit into mined blocks next 24h.
reply
Exactly my point: assuming no new transactions come in is ridiculous. New transactions always come in. And in circumstances where there's been high enough demand to reach the mempool limit, it's even more likely that new transactions will be coming in!
reply
5 days uncomfirmed with a 1sat/vbyte, to myself so no worries, more of an experiment. Just bumped the fee to 3sat/vbyte and the default rate went up more for the current block. I'm in no rush. But it is fun watching all this. Happy for the miners during the current market.
Also all this is showing me how important Lightning is for instant and low fee transactions. I got too accustomed to empty mempool layer one transactions. Like all things. This too shall pass.
reply