reply on: Why I'm Against Covenants (Jimmy Song) \ stacker news ~bitcoin

pull down to refresh

100 sats \ 11 replies \ @SimpleStacker 12 Sep \ parent \ on: Why I'm Against Covenants (Jimmy Song) bitcoin

UTXO set bloat is one of them. IBD on same hardware went from 3 days to 1 month for me. There is some discussion here that I was a part of: https://github.com/raspibolt/raspibolt/issues/1482#issuecomment-3263903647, to show that it's being experienced by multiple people.

One could say, get better hardware, but I thought one principle was trying to keep the cost of node running down.

30 sats \ 7 replies \ @optimism 12 Sep

what are you writing the block data to?

100 sats \ 6 replies \ @SimpleStacker 12 Sep

https://www.amazon.com/dp/B0874Y5XFG

90 sats \ 5 replies \ @optimism 12 Sep

That's odd. I have plenty of these, so the disk is not an issue. I used it on an older Rockchip board (comparable to Pi4 except it has sha256/crypto extension) and it took 4 days back in April. I'll try to source me a Pi4 next week - worth looking into.

100 sats \ 4 replies \ @SimpleStacker 12 Sep

Thanks. The github discussion I linked above (copied again for convenience) shows that multiple users are experiencing this. My hypothesis was that the UTXO got too large to hold in RAM and so it is using HD, but I could easily be wrong.

https://github.com/raspibolt/raspibolt/issues/1482#issuecomment-3263903647

30 sats \ 0 replies \ @optimism 13 Oct

Have charted debug=bench on my RPi4 8GB w/ 2TB Samsung T7 (and txindex=1 per raspibolt recommendation) for the first 240k blocks as a test and to develop some tooling. The x axis is 1k blocks cumulative.

This means that my log parser works so I will now plan an uninterrupted session when I don't have to fly in the middle of the week, to chart it all out completely. When I know what blocks to look at, I can write a custom bench-bitcoin routine so we can trace it on different platforms and see what's what.

0 sats \ 2 replies \ @optimism 12 Sep

Saw your comment on there previously. I haven't tried raspibolt personally and there is some ongoing work w/ core towards this - I must admit I used 27.x, and there are reports of a regression with 29.x: see https://github.com/bitcoin/bitcoin/pull/32043 and https://github.com/bitcoin/bitcoin/issues/32832

Shall I get a 4GB RPi4?

100 sats \ 1 reply \ @SimpleStacker 12 Sep

Interesting about the regression. That's up to you! I currently don't have any new insights, and my node is finally fully synced.

0 sats \ 0 replies \ @optimism 12 Sep

Never mind, I didn't read well and you have an 8GB - I'll try to get that.

Edit: will get it next week.

0 sats \ 2 replies \ @ek 12 Sep

Okay, fair, I forgot about this!

100 sats \ 1 reply \ @SimpleStacker 12 Sep

I think it's easy to forget because it's not a big problem if you're already running a node that's been up for a long time, since you only need to validate once every 10 minutes or so. I only realized it when I decided to rebuild my node from scratch.

100 sats \ 0 replies \ @lightcoin 12 Sep

it is worth noting that this same "problem" would also occur if bitcoin had a large number of UTXOs for other reasons (like lots of users taking self-custody)

it's a problem that would need to be solved (and is being solved by solutions like zerosync and utreexo) regardless of the use-case that caused it.