log☇︎
2200+ entries in 0.199s
mircea_popescu: did the orphanage burner ruin trb 's chances of unwedging in this situation ?
mircea_popescu: yeah. and since you mention it, trb-i definitely needs a clarified push-or-pull model because the current system is the soul of unconsidered adhocery
asciilifeform: http://btc.yt/lxr/satoshi/source/src/main.cpp?v=makefiles#1364 and http://btc.yt/lxr/satoshi/source/src/main.cpp?v=makefiles#1735 are the only times a trb node asks for blocks explicitly from peer ☟︎
mircea_popescu: no argument there ; you however may in turn recall that trb is by inheritance an utterly chtonian horror of heap allocation etc.
asciilifeform: mircea_popescu: you might recall that trb does not attempt to decode tx scripts (yes) when verifying block
mircea_popescu: asciilifeform no, i'll complain to you, because really there's no need trb logging be THIS RETARDED
asciilifeform: mircea_popescu: if not mega-seekrit, did this node peer with well-known trb nodes (e.g., mine ) ?
mircea_popescu: it'd be tremendously helpful for instance if the trb node had found it within its good graces TO FUCKING PUT TIMESTAMPS IN THE LOG.
asciilifeform: mircea_popescu: a block doesn't get to sit down in blk**** in trb unless its antecedents are present.
ben_vulpes: 0036.dat also implies a trb .dat file, not a dumpblocked file? happy to crack it open and see though
asciilifeform: pete_dushenski: mircea_popescu showed up with a bag of clues from a trb node that wedged some weeks ago. so now autopsy.
asciilifeform: incidentally trb probably ought to shit a sha512 of incoming block into the debug log
asciilifeform: trb's block push/pull mechanism is so retarded, that it is possible for a node to go for eons in a wedge, simply from never receiving the necessary unwedge blocks.
mircea_popescu: and inasmuch as node wasn't capable of extracting itself naturally, and it IS a trb node, this qualifies as successful attack against network, by and large.
asciilifeform: does it still, for instance, have orphanages ?! because that ain't really modern trb, in any real sense
mircea_popescu: was not, it's an older trb.
mircea_popescu: in other lulz, i found a trb node which is locked on block 419373 and dumps all blocks as unacceptable bastards
PeterL: while we are talking about things to stick in TRB-I, how about lowering the block size by an order of magnitude or so?
a111: Logged on 2017-02-27 16:56 mircea_popescu: but the correct trb-i might just as well end up this situation where block reward is 1mn bitcoin, and it dies within 1mn blocks. so all mining does is produce ~ a lease ~ on a chunk of bitcoin. and the value of old bitcoin is monotonically decreasing over their lifetime.
mircea_popescu: this item definitely counts for your grand list of trb-isms. on the strength of that, "computable", i ask no more.
mod6: yeah, trying to keep up with all these posts. just started this one on "possible trb-i"
mod6: <+BingoBoingo> <mircea_popescu> trb-tits << I thought that was shinohai's fork << :D
BingoBoingo: <mircea_popescu> trb-tits << I thought that was shinohai's fork
mircea_popescu: trb-tits
BingoBoingo: asciilifeform: In my mind trb-i is discussion piece. Ideal may be impossible. trb-i may end up being way to consider idea for trb-b which suceeds trb-a (a is for arse)
asciilifeform: BingoBoingo: keep in mind, it's ~a possible~ trb-i, not ~the~... thing's quite certainly still in 'matrix mechanics' stage of life, and who knows, for how long.
BingoBoingo: Oh the Trump flensing knife lulz come out tonight, while still digesting trb-i theory. What a time to be live!
asciilifeform: mircea_popescu: we had this thread, trb retains orphaned branches of any and every size, because it never goes back to remove anything from ~inside~ a blkxxxx.dat .
asciilifeform: ^ if anyone recalls the 'eatblock' thread -- i found that my blkxxxx.dat differed, in a handful of places, from mircea_popescu's, and yet again from every other trb node's.
mircea_popescu: it's not altogether clear to me how such a thing is an improvement over "just run your current trb through the future gossipd"
asciilifeform: the fixed-width-tx is a provable component of any long-term-sane trb-i. regardless of what other parts are included or excluded. without it, you get rapid rot.
asciilifeform: asciilifeform's aim was to consider gedankexperiment trb-i where this cock is turned around radially, at the miner. who, reaping most of the cake, ought to also absorb the cock.
asciilifeform: [BTC-dev] (CRACKPOTTERY) Notes re: one possible "TRB-I". ☟︎
asciilifeform thinks 'holy fuck is the trb-i article looooong;' who will have the strength, to read this.
asciilifeform writing draft trb-i spec
asciilifeform: http://btcbase.org/log/2017-02-27#1619482 << right now, keeping a node is -ev for almost everyone who could be doing it. only oddballs with countereconomic motivation of one kind or another (e.g., trb experimenters) , plus miners themselves, plus serious txers ( e.g., mircea_popescu ) have a desire to do it. there are not so many of these. it is rather like relying on entirely on coprophagics for your sewage disposal needs. ☝︎
asciilifeform: trinque: say we stick to the trb-i thread. gotta specify what specifically about your concept of trbi, that would remove the incentive for miner secrecy that exists in classical bitcoin.
asciilifeform: trinque: this is doable right now, you can comment out the mempool in trb...
mircea_popescu: but the correct trb-i might just as well end up this situation where block reward is 1mn bitcoin, and it dies within 1mn blocks. so all mining does is produce ~ a lease ~ on a chunk of bitcoin. and the value of old bitcoin is monotonically decreasing over their lifetime. ☟︎☟︎
asciilifeform: mircea_popescu: just to make it absolutely clear, i don't see a long-term future for satoshi's turd. all of my work on trb is to be regarded in same light as the neutron-absorbing armour on 1970s sov tanks -- something with which to prolong the life of the crew ~slightly~ so that it can drive over freshly-nuked ground and last a few hours of shootout.
asciilifeform: to revisit much further upstack, to http://btcbase.org/log/2015-02-14#1018732 ( via mircea_popescu's latest article ) -- consider a 'trb-i' where a tx carries proof of work, and is likewise mined as is the block ☝︎
mircea_popescu: asciilifeform yes trb does reindex.
jurov: and can trb live with terabyte tx index at all?
asciilifeform: i suppose one way to rebuild the index using existing mainline trb would be to nuke the .bitcoin dir entirely, and refeed the node via 'eatblock'.
asciilifeform: trb does not have a knob for 'reindex and make sure blkxxxx matches the index'.
asciilifeform: now you need extra logic in trb per se, and this is no longer 'cheap perl hack'.
asciilifeform: well not entire fs, but every file touched by trb
asciilifeform: very concretely: all enemy has to do, to gum up a trb node, is to repeatedly request blocks from 0 .. current, randomly, and switch ips as needed
asciilifeform: (a trb-i item )
a111: Logged on 2016-12-29 23:20 asciilifeform: type2 ( pete_dushenski's ) is the garden variety shitflood. which is sometimes solved by ip ban, but only in the case of 'shrapnel addressed to occupant', i.e. idiot prb nodes wildly spamming crapolade, and not in the 'bullet with your name on it' case, where somebody actually has a sybil constellation drowning your trb node in liquishit, with no SINGLE ip misbehaving in any way
deedbot: http://trilema.com/2017/trb-i-addressing-scheme-proposal/ << Trilema - TRB-I Addressing Scheme Proposal
asciilifeform: while we're doing trb-i : in addition to 'tx is 1024 bytes, and block is 1024 tx' , consider another item: 'block MUST contain 1024 valid tx'
a111: Logged on 2017-01-17 00:21 asciilifeform: to possibly squeeze something useful from thread: as i understand, a lamport-based 'trb-i' ~could~ run on z80.
mircea_popescu: absent a good or at least workable breakthrough in this vein, there's no strong technological incentive to move to trb-i
mircea_popescu: idiot example #2 : a trb which allows txn to be blocked by others than their issuers is ALSO a "way to do things" which doesn't in fact work, and therefore, exactly equivalent to the peter todd & prb idiots item
mircea_popescu: yes, but the idea is to not expand the hipster doofus design principles to trb-i
mircea_popescu: the g has a decent debottler built in ; the trb-i does not, and needs a few.
asciilifeform: in given trb-i scheme, a block either exists on disk -- or does not
asciilifeform: locking problems (gotta add tx to index, in current trb, as aggregate -- but the only way to do that is to stop the world! like complete idiot -- every time there is a new block, to prevent situation where there is a partially indexed block available to incoming mempool tx verifier) disappear.
asciilifeform: in either hypothetical trb-i -- you no longer need a db. any db.
asciilifeform: to clarify -- in my mind, a 'trb-i' ~must~ be capable of checking validity of tx at wire speed (i.e. at the speed it is physically capable of receiving them) on reasonable iron.
mircea_popescu: trb-i, yes?
asciilifeform: trb node is, for instance, continuously engaged in the sin of 'something to allcomers'.
asciilifeform: ben_vulpes: pretty idle lately , actually, trb is a procrastinatory escape from the hardstuff
jurov: https://bitnodes.21.co/nodes/?q=Slovakia i finally put up permanent trb node (it's the syncing one), but have nfi about the other two
mircea_popescu: anyway, asciilifeform, you should prolly publish a codebase adnotated with time profiles. trb-454523 ; trb-454520 etc.
asciilifeform: would seem that these are a type of pseudonode / misc. attacker, that comes in two varieties, one aimed at recent prb (majority), the other -- more trb / old-prb - flavoured.
asciilifeform: in other lulz, there is a large population of nodes reporting 'version message: version 60000, blocks=350000' for all eternity (typically they auto-disconnect when discovering trb ver.) . anyone know who they belong to, and wtf ?
lobbes: http://btcbase.org/log/2016-03-03#1421109 << btw, thank you for this, alf. I will be embarking on my own gentoo quest soon to finally stand up a trb node ☝︎
asciilifeform: (stock trb will happily drop ~anyone~ on the floor, for dozen different reasons, incl. 'we used him for too long')
asciilifeform: unbitflippable direct pipe to large trb node.
asciilifeform: unrelatedly, system of two 'wired' trb nodes appears to have marked resistance to 'blackhole'.
asciilifeform: adlai: any time you feel like doing something useful -- no shortage of items. there is, for instance, a trb knob direly needing testing, http://therealbitcoin.org/ml/btc-dev/2017-February/000251.html
asciilifeform: ( if you recall , that was when we had a trb node, offline, eat the complete chain to date from disk, with timer and valgrind going )
asciilifeform: to what else would you have trb 'respond as they expect', PeterL ? bloom filter? segshitness?
asciilifeform: classical trb did not have a nondisconnectable marker for peers; would not connect to localhost; and deprioritized non-8333 ports to dead-last order. 'wires' fixes this.
shinohai likes to minagine this `wires` patch as G for trb
a111: Logged on 2016-01-20 02:05 asciilifeform: presently trb does not have this sane behaviour
asciilifeform: in other noose, http://wotpaste.cascadianhacker.com/pastes/jJ7iS/?raw=true << on a traditional (zoolag) trb node. i never once saw these in past log readings, and now they are quite common.
asciilifeform: considering that it cuts off stock trb -- it is surprising that it was at any point caught up.
asciilifeform: (4) is unfortunately ultimately rubbish, it is really the single-threaded, polled socket handler of trb that is ultimately morally responsible.
asciilifeform: 1. 'blackhole.' 2. tcpdump on two blackholed trb nodes. multitude of peers emitting 'ping, ping, ping...' and soaking up sockets. 3. hypothesis: killing socket hoggers will dissolve blackhole. 4. 'socket-hogging prb is responsible for blackhole condition'
asciilifeform: to briefly revisit upstack: in case this was unclear to anyone: goodbye-pingers ~breaks connectivity with stock trb~ and ought to be considered experimental/dangerous . it is not yet clear to me that the pill is an improvement over the disease.
asciilifeform: because as it is, trb nodes (such as zoolag) are getting the hammer.
asciilifeform: (or rather, no longer emits pings. trb nodes emitting pings -- still banhammered)
asciilifeform: this one, lel, no longer bans trb nodes...
asciilifeform: in other lulz, 104.199.165.17 << google spider . now attempts to connect to trb nodez.
asciilifeform: (see the logs of any operating trb node)
asciilifeform: i.e. 'trb-i'
asciilifeform: right now this is not even a very interesting wtf. it may however become interesting if a miner decides to sit down on trb island.
asciilifeform: my concern is not that there is a boojum in 454074. but that the thin red line connecting my trb nodez to miners are -- apparently -- veeeery thin
mircea_popescu: asciilifeform> to continue: what portion of cpu time is spent verifying ultimately invalid blocks? <<< yes, yes and yes, altogether a proper STATs module should very much be part of stock trb, and very helpful if it were.
BingoBoingo: asciilifeform: Bastard 0.7 ish node corrected will patch trb nodes at convenient time, tyvm
ben_vulpes: it should be doable to 'getpeers' from the nodes from which a trb gets a block and then attempt to traverse upstream, no?
asciilifeform: or, how much time does trb node spend rejecting crapolade from same idiots again and again
mircea_popescu: originally, miner code got split off the main satoshi base sometime in 2011. there was a lot of back and forth between pre-trb prb node code and the node code miners used, but not that much in the past few years.
mircea_popescu: at that point, the writhing horror (which you think of as prb) had about 10x as many loc as trb does ; by now it's 100x.
mircea_popescu: asciilifeform it drops them. and i am pretty clearly remembering we talking about this in the early days of trb and foundation
asciilifeform: box falls behind, loses peers, begins to sink into classical blackhole behaviour (as it gets dropped by everybody, including fellow trb nodes)
asciilifeform: so they see trb set to, e.g., 99999 (as mine, for instance, are) and goes ping,ping,ping,motherfucker,ping,ping,whywontchapong