233000+ entries in 0.146s
ben_vulpes: all of a sudden i want
to collect data on how long
trb
takes
to return from a simple getinfo call, per your above protocol
ben_vulpes: i may also need
to adjust my log rotator, 1 gig of bitcoind logs is...not actually
that much history.
mircea_popescu: i suspect it can be done by surgery without
the
two parts ~even knowing~
they're not satoshi-full-bitcoin.
mircea_popescu: and you know, sanity suddenly flowers everywhere - do you send
too much in bytes/s ? b.peer can can your. do you send
too much crap ? b.blockchain can
tell b.peer it dun wanna hear from you no mo.
mircea_popescu: anyway.
the divorce is required and continuing in
the current monolith sheer nonsense.
mircea_popescu: if queried, b.peer loads mempool-txn as it is and uses
that ; so what if it's stale, fuck you.
mircea_popescu: b.blockchain picks up items from received-blocks and verifies
them, adding
to blockchain if necessary ; picks up items from received-txn, and verifies
them, adding
to mempool-txn if necessary.
mircea_popescu: b.peer receives items from peers and puts
them in received-txn and received-blocks ; both of which are queues.
mircea_popescu: asciilifeform it answers immediately. let's formalize
this.
tb0t: Project:
trb, ID: 35,
Type: I, Subject: Investigate blackhole, Antecedents: , Notes: Investigate what might be occuring with
the so-called black-hole, described here: btcbase.org/log/2016-12-20#1586635
mod6: !%a
trb I "Investigate blackhole" "Investigate what might be occuring with
the so-called black-hole, described here: btcbase.org/log/2016-12-20#1586635"
mircea_popescu: business has a secretary and an actual worker ; fast food has a front office and a back office.
there's reasons for
things wtf.
mircea_popescu: the fact
that you need
the blockchain part
to
tell you what
tx are
to be relayed does not reduce
to, you must make it
to listen
to all comers.
mircea_popescu: nope. blockchain part will get
to it when it gets
to it, and
tell you. until
then, peer part builds queues.
☟︎ mircea_popescu: mixing
these into a single frail monolith is very much satoshism, but we're not supposed
to stick
to stupid.
mircea_popescu: the process
that
talks
to peers has ~no business~ knowing anything about
the blocks ;
the process which maintains
the blockchain, verifies blocks etc has ~no business~
talking
to peers.
mircea_popescu: at
the very least block digestion and peering must be cleaved in
trb
☟︎ mircea_popescu: the miners are pushing block complexity
to
the maximum possible because hey, more fees ;
the cost
this imposes on
the nodes is not
to be discussed because hey, fatso
thinks he should matter and stuff.
mircea_popescu: whole story is, "fatso is angry at not being as relevant as he
thinks he should be"
mircea_popescu: practically speaking,
the exact equivalent of "putin did wtc"
mircea_popescu: this butressed on
the observation
that satoshi code is large and stupidly wrought.
mircea_popescu: asciilifeform
the suspicion is you'll be seeing
too much variance
to be able
to say much definitively
mircea_popescu: myeah. i have no problem with
the baruch spinoza approach, esp when eg, dealing with viri etc. but it's not a universal wrench.
mircea_popescu: imo
this is a
total waste of
time ; we won't have
the culprit even if we drink
the gigabyte swamp. which we needn't be drinking in any case.
mircea_popescu: just keeping
track of
the binary star is already a huge
thing.
mod6: oh
that's not it
then asciilifeform?
this is helpful, please continue
to describe.
ben_vulpes: during which
the node is ~entirely unresponsive
ben_vulpes: mod6: no,
the
thing where either a shitty network client or some joker opens sockets and lets
them expire, eating 60 seconds of
the loop
through each peer
to service
ben_vulpes: i don't expect
to be able
to make an rpc call while
the node is looping
through its list of peers and get 'the right' peer shat out
mod6: im not sure
that is helpful... are you discussing how
the client just seems
to have no peers after some period of
time?
ben_vulpes: asciilifeform: i mean when
the node is running
through 'accepted connection' 'socket no message in first 60 seconds'
ben_vulpes: might be interesting
to patch
trb
to dump relevant connection's self-identification string
mircea_popescu: to
think
the dork actually claimed he is "slightly more productive on linux
than windows". satoshi never fucking as much as saw a posix compliant box.
mircea_popescu: nevertheless, bitcoin 2.0 has no room for
this serial nonsense.
mircea_popescu: anyway. like you said, seems pretty credible parallelizing will fix
this./
mircea_popescu: yes but
this doesn't answer if
the idiocy is naturally ocurring so much
a111: Logged on 2016-12-20 19:04 mircea_popescu: we have no definitive answer on
this.
ben_vulpes: but i mean blackholing as artifact of some other poorly written client, instead of script opening sockets
to
trb nodes
mircea_popescu: but in general
there is no such
thing as an unwanted pregnancy. for
the attack
to exist,
the client has
to be "retarded".
ben_vulpes: is
there intuition about whether blackholes are attacks or retarded clients?