15:00:20
rucknium:
MRL meeting in this room in two hours.
16:59:47
rucknium:
Meeting time! https://github.com/monero-project/meta/issues/1318
16:59:53
rucknium:
1. Greetings
17:00:04
vtnerd:
Hi!
17:00:12
rbrunner:
Hello
17:00:25
articmine:
Hi
17:00:30
ArticMine:
hi via IRC
17:02:11
rucknium:
2. Updates. What is everyone working on?
17:03:06
rucknium:
me: Working on some updates to https://moneronet.info/ to properly display new spy node behavior.
17:03:35
rbrunner:
In the middle of the biggest review I have ever done: sneedlewoods' switching `simplewallet` from directly using `wallet2` to using the Wallet API
17:03:48
articmine:
I updated the scaling documents to fix the wallet scaling to 1.2 on the long-term median.
17:04:09
articmine:
Also attended the 39c3
17:04:34
rbrunner:
Oh!
17:06:36
vtnerd:
Me: debugging an otherwise finished fcmp++ tx construction in lwsf
17:07:15
rucknium:
3. Spy nodes (https://github.com/monero-project/meta/issues/1124).
17:08:59
rucknium:
I have been working on a few improvements to https://moneronet.info/ . In the current version, a node is only counted as spy if it behaves like a spy node in the daily scan. But I want to change that to also include nodes on the (new) MRL ban list. There are many spy nodes on Hetzner and Digital Ocean that stopped having the spy node fingerprint, but are probably still spy nodes.
17:10:29
rucknium:
The webapp is showing a decline in the number of honest nodes, but I think it's a false reading.
17:11:34
rbrunner:
What's your theory how you get it wrong now?
17:11:56
rucknium:
I think a problem with database locking.
17:12:21
rbrunner:
So just some problem inside your app?
17:12:26
rucknium:
I don't know why it's causing a noticeable problem now and not before.
17:12:58
rucknium:
Yes. I think so. I ran @boog900:monero.social 's simpler scanner and it picked up the expected, larger, number of nodes.
17:13:28
rbrunner:
No visible decline because of the holidays? I would expect at least some decline.
17:13:40
rucknium:
My scanner also saves peer lists and I think everything else a node sends when it reponds to a Levin handshake.
17:14:13
rucknium:
So I need a database for it instead of just writing to a flat file like the original scanner did.
17:14:48
rucknium:
Well, that trend, if it exists, is the victim of lost data 😬
17:15:49
rucknium:
I want to have this ready for a social media announcement of the new version of the MRL ban list next week.
17:16:03
rucknium:
IMHO, not a good idea to announce it during this week.
17:16:22
rbrunner:
Yes, too quiet
17:16:38
rucknium:
I plan to have two lines for the count of the nodes that have the MRL ban list. One for version 1 and one for version 2.
17:16:54
articmine:
Yes very quiet
17:17:40
rucknium:
selsta also asked rbrunner and me if we think the ASmap is worth deploying: https://github.com/monero-project/monero/pull/7935#issuecomment-3666660926
17:18:29
rucknium:
I plan to reply no because subnets are much less risky to use as a basis for selecting peers.
17:18:38
rbrunner:
No idea about that "ASmap" stuff, never looked into it
17:18:56
rucknium:
Subnets always have the same number of possible IP addresses. ASNs do not.
17:20:18
rucknium:
ASNs are more easily manipulated by adversaries. According to @boog900:monero.social , the ASN that the spy nodes are now using, instead of LionLink, is only one year old. Nodes would have to constantly update the ASN database or they would be victimized by such an adversary, IMHO.
17:20:54
rbrunner:
Some easy automated lookup is not possible, or at least complicated?
17:21:23
rucknium:
I have a game theory model that shows pretty rigorously the conditions where having subnet deduplication is better than not having it. It's based on knowing the subnet size, i.e. number of IP addresses in eash. ASN size can be tiny or huge.
17:22:22
rbrunner:
Anyway, it would probably complicate things even more, for unclear gain, or hard-to-prove gain
17:22:40
rucknium:
Game theory models don't map perfectly onto reality, of course, but at least it does inform about worst case scenarios IMHO.
17:23:37
vtnerd:
generating the map is automated but is reliant on your view of the ip peering protocols
17:23:37
rucknium:
Game theory model is here: https://github.com/Rucknium/misc-research/blob/main/Monero-Peer-Subnet-Deduplication/pdf/monero-peer-subnet-deduplication.pdf
17:23:38
vtnerd:
https://asmap.org/sourcing-data/
17:23:41
rucknium:
Not very complicated IMHO. Two-player zero-sum game with two possible pure strategies each.
17:24:04
vtnerd:
I’ve never liked it because the db changes how the connections work, and the db itself is hard to review
17:24:31
rucknium:
Isn't the proposed BTC implementation a static database that ships with the node binary?
17:25:30
vtnerd:
yes. its just that you’ve got to more or less trust how the db was made. or I guess you could review all of that generation code and then attempt to rebuild the db yourself
17:25:56
vtnerd:
the process just seemed to opaque for more liking, but other people in btc seemed to disagree
17:27:22
rucknium:
It's not merged in bitcoind yet, right?
17:28:23
rucknium:
AFAIK, the purpose of the bitcoin implementation is a little different, or different emphasis. Their emphasis is anti-censorship and network resilience, not really privacy AFAIK.
17:29:32
rucknium:
Anything more on spy nodes?
17:30:13
vtnerd:
not sure of the merge status, looks like some portion of it may have been merged (the code but not the db)
17:30:28
rucknium:
Is @preland:monero.social here?
17:31:00
vtnerd:
yes they are more concerned with eclipse attacks, etc
17:31:40
rbrunner:
Much more money involved. Imagine you could cut of some important miner, for even a few hours ...
17:33:41
rucknium:
I will skip the "Post-FCMP scaling concepts" item this meeting since preland isn't here.
17:33:50
rucknium:
5. FCMP alpha stressnet (https://monero.town/post/6763165).
17:34:20
rucknium:
Bugs are still being reported and squashed
17:34:49
rbrunner:
Running already for 3 months, if the date of the post is correct. Time flies.
17:35:01
rucknium:
This alpha stressnet has continued longer than I expected, but it's continuing to be useful
17:35:46
articmine:
How big is the block weight?
17:36:37
rucknium:
I remember telling @untraceable:monero.social that the stressnet blockchain would go to about 50GB. I don't know how big an unpruned node is now since I pruned all my nodes. But it's above 150GB AFAIK :D
17:37:30
rbrunner:
Uh, and just now SSDs prices will probably explode :)
17:37:36
articmine:
What about ML?
17:37:43
articmine:
MP?
17:38:48
DataHoarder:
due to orphan forks the long term weight wasn't increased that much afaik
17:39:14
DataHoarder:
and stuff breaking down after 15-18 MiB and nodes falling out of sync -> making their own empty blocks, then merging later down the line
17:39:15
articmine:
Sorry MN the former MS
17:40:03
articmine:
So MN resets