asciilifeform: potentially could sharpen an already dire 'ecological' problem.☟︎
asciilifeform: Mocky: the current algo optimizes for nuffin at all. but the puzzler is, if we were to optimize for ~something~, could it create a problem. esp if the algo worx as i predict, and the nodes using it, end up forming an optimally short link to the miners.
asciilifeform: but in general blox come from asiclandia.
asciilifeform: 'who' is a misconceived, somewhat, q, cockroaches dun have faces
asciilifeform: Mocky: recall the chains of empty/near-empty blox the chinese sometimes emit
asciilifeform: Mocky: the boojum is that mining in classical btc worx in such a way as to amplify small advantages into gigantic
asciilifeform: but on the other hand, the current system makes for pestilential sibylism.
asciilifeform: this is a potentially dangerous arse-mouth system.☟︎
asciilifeform: and not merely because A,B,C reliably supply correct blox, but they also will have most reliable tx-eating. so they will get sent all new tx in preference to D.
asciilifeform: thereby any peer ranking algo that favours 'useful' peers, will give A,B,C a substantial advantage over any new D.
asciilifeform: gedankenexperiment : let's suppose that all new blox presently are born in nodes A, B, C. from there, propagate to rest of planet. the immediate peers of A, B, C, will have them parked at the very top of their peer table, right below the manually-added 'meat' peers. now say a D shows up. D may have faster hash rate even than A+B+C, but his blox will still be considered only ~after~ those of A,B,C. at least, for a time.
asciilifeform: or, rather, slightly increases the 'inertia'
asciilifeform: ( potentially such network may favour 'old' miner over 'new' )
asciilifeform: i'd be interested to see what mircea_popescu makes of this algo. because i can see a possible fatal downside.
asciilifeform: ( nao, ideally all of the sane people would actually know one another and wot, and link their nodes via ciphered pipes, all the way to miner. but i dun expect to live to see this. )
asciilifeform: whatever machine cycles ~remain~ after this, can service randos.
asciilifeform: sanely-behaving nodes (both trb and hypothetical trb-compatible heathens, which exist, or we'd see 0 blox) ought to be able to find ~and keep~ each other company
asciilifeform: arguably 1st step to properly de-sibyl-ing the network.
asciilifeform: i suspect that it would lubricate both block rx and tx broadcast.
asciilifeform: this will make for something like sane peer propagation, which presently we dun have.
asciilifeform: imho a trb node ought to only advertise addrs that either a) part of the manual --addnode set the node was brought up with , or b) actually supplied a correct block in the recent past
asciilifeform: mod6: right nao, our address-getting mechanism is the ancient turdalicious one inherited from shitoshi -- where the table is blindly filled with whatever connected folx decide to throw in
asciilifeform: mod6: my main hesitation re the item is that it may be 'plugging wrong end of funnel', quite possibly no such gymnastics would be necessary if we simply had a http://btcbase.org/log/2018-09-23#1853143 and always --addnode'd the actually useful nodes, rather than relying on pure chance☝︎☟︎
asciilifeform: mod6: i was thinking of trinque's idea : suppose trb closed all open pipes if it finds that $configurable hours (e.g. 3) have passed without new blox
asciilifeform: diana_coman: traceroute --mtu destinationip will show the mtu of the 1st 'fraggable' node in the path☟︎
asciilifeform: e receiver’s reassembly timer. For higher delay high capacity network paths, this limit of 65,535 packets in flight can be a potential performance bottleneck [RFC 4963].'
asciilifeform: in particular, 'Lost fragments represent a slightly more involved problem than lost packets. The receiver has a packet reassembly timer upon the receipt of the first fragment, and will continue to hold this reassembly state for the reassembly time. The reassembly timer is a factor in the maximal count of packets in flight, as the packet identifier cannot be recycled within the period defined by the sender-received path delay, plus th
asciilifeform: ( this braindamaged design decision goes, as one might expect, back to the olden days when homo redditus was not yet on the net and 'arpa' designed for nukefest-related problems, not against 'ddos' )☟︎
asciilifeform: the other item of interest, not mentioned, is that quite a few infrastructural routers will simply drop frags, for the reason that all but 1 frag of a fragged packet contain no copy of the source or destination port -- and thereby impossible to ddos-filter. so they drop'em.
asciilifeform: ( can skip straight to section 'IP Fragmentation' )
asciilifeform: the 1 possibly noteworthy bit, is that fragging can take place anywhere in the route, while reassembly -- only at final receiver .
asciilifeform: diana_coman: https://archive.is/k6Wxb << udpism likbez ( may or may not contain anyffing new to you, but posting for the l0gz/n00bz )
asciilifeform: ave1: i was speaking generally, 'are there other hardwired pathisms in there'
asciilifeform: the other simple experiment, re fragging, is to set up the receiver on local host, and then on same-lan box
asciilifeform: diana_coman: do you have a list of the ones that reached the other end ?
asciilifeform: diana_coman: tcpdump -i eth0 udp port YOURPORT -vv -X
asciilifeform: ( i long suspected that nothing far over 512, will reliably go )☟︎
asciilifeform: diana_coman: the other thing, entirely possible that yours got fragged and never unfragged, this'll be path-dependent ( as discussed in orig thread )
asciilifeform: according to the man pg, sento() block unless one were to set MSG_DONTWAIT ( and we didn't )
asciilifeform: diana_coman: what was your random size range ?
asciilifeform: diana_coman: unless i misread the docs -- yes
asciilifeform: afaik not if you're using blocking send ( and my routine, does )
asciilifeform: diana_coman: iirc max packet that linux will actually send, is 65507 byte
asciilifeform: !Q later tell BingoBoingo crate seems to be in customs limbo still
asciilifeform: ideally in the end there will be none.
asciilifeform: ave1: pretty neat. curious, didja grep the whole thing for hardwired paths ?☟︎
asciilifeform: btw ( and this is not mega-revelation... ) 99.999..+% of noad traffic, is worse than useless -- box is speaking with noads who do not either bring it luscious next valid blox, NOR carry yer tx'en closer to miner, NOR even serve up full chain and enforce trad rulez thereon
asciilifeform: it's a network problem, not a trb one.
asciilifeform: trinque: not long ago noticed a 400 blox gap on mine, where interestingly restarting did not fix, had to sit for the typical n days and finally meet noads who actually have new blox and dun get malleus'd
asciilifeform: i for one have my kernel, and dun expect to ever miss the Official 'new'
asciilifeform: homo bigmackicus will vote for whatever tv instructs, yes
asciilifeform: reminds of old-school 'grifters', the kind that marry 85yo heiress
asciilifeform: and from this, goes to have power of attourney and trademark rights? bit of a climb eh
asciilifeform: i still dun fully grasp the process whereby a theo 'marries' a bobbeck. what's the draw, exactly
asciilifeform: mircea_popescu: i suspect, in these, simple cowardice, they aimt eager to move out of the mental universe where 'law and order' and into the lafond planet where earthly carcass actually finds itself
