449900+ entries in 0.276s
mircea_popescu: <jurov> otherwise either kernel stalls or pipe is congested(dropped packets, whatever) << i
thought
this may well be it previously, as load was uncharacteristically high for
this application. but solved
that problem and apparently no improvement.
decimation: the stereotypical 'buffer bloat' problem is
that you find packets become extremely lagged when
there is heavy bandwidth use
decimation: assuming
that
this is
the last machine
that
touches it before it goes
to your isp
jurov: with general conclusion
that if SendQ buffer is empty
then it's application fault
decimation: yeah, find ascii's connection and see what
the sendQ value is
mircea_popescu is
trying
to bootstrap
the public interest in
the matter into getting a networking education for himself.
mircea_popescu: essentially
the adult version of "but daddy, all
the kids in class got
this question wrong
too!"
mircea_popescu: because yes,
the point of life is
to preserve
that most valuable delusion -
that "nobody could have foreseen..."
decimation: what's amusing is
that nobody gets it right, or apparently even
tries
mircea_popescu: i'd withdraw anything i had in
that shithole by now, had i not withdrawn it long before, had i had anything in
there in
the first place.
mircea_popescu: <decimation> or why
those filthy laptops can come anywhere near
their 'hot wallet key' <<
this is easy enough
to answer, if you
think about it. and if you don't... if you let history be your guide.
mircea_popescu: so
they spent 250k (abnd counting) so as
to lose 4mn off a schmuck's home connection ?
decimation: yeah, but if each pawn is
too week
to carry weapon, have nothing
ascii_field: decimation: pogo is
to have strength in numbers
ascii_field: mircea_popescu: connects, disconnects, without
transfer of anything
decimation: ascii_field: note
the statistics from backblaze don't show
the probability of single bit errors (undetected by
the drive)
decimation: or why
those filthy laptops can come anywhere near
their 'hot wallet key'
ascii_field: tickets
to Punk Rock Holiday 2015. (Merlak is keen on punk rock and has played in a band.)'
ascii_field: 'The gambit for
this phishing attack was
to offer Mr Merlak free
ascii_field: 'In addition,
to prevent future capital losses of
this kind, we have contracted with a vendor
to provide “multi-sig”
technology
to better protect our hot wallet (this particular
transfer could not have happened
today) and hired a skilled
technology company, Xapo,
to assist in managing our cold wallet.'
ascii_field: 'In addition, we have paid out approximately $250,000
to programmers hired
to rebuild and improve our platform; paid approximately $250,000 (and counting)
to
the Stroz Friedberg
team; and at least $150,000 more for various security reviews, and legal and financial advice.
These out of pocket costs are continuing
to accrue.'
jurov: well,
that would show in netstat output?
ascii_field: i'm gonna guess
the entire link between anywhere i can go without
taking a plane, and mircea_popescu's box, is burpy on account of packets hitting
tape before
they get
to me
ascii_field: after
that, red phone in derpistan dept. of derp rang
ascii_field: mircea_popescu announced
the box ip publicly
decimation: if you are using a standard linux distro, it's well known
that older kernels had buffer bloat/tcp
tuning issues
jurov: if ssh works fine, you'd confirm
that someone is doing packet inspection
mircea_popescu: decimation yes, but
this server would not stand if
that were
the case.
mircea_popescu: we know we can actually pass data between each other. what we wish
to find out is what causes
this behaviour and who
to kill.
ascii_field: ssh
tunnel is what i suggested earlier aha
ascii_field: where
the buggers use what
they have - control over backbones -
to annoy people
jurov: try
to run it in ssh
tunnel?
ascii_field: i suspect we have come
to a 'chinese' point
mircea_popescu: ascii_field nuts. i
tell you ...
the load is < .2 and has not peaked above 1 (this is on 8 cpu machine).
there is nothing deferred or anthing suspicious in
the logs.
the eth card is doing fine.
two dozen connections, routinely hitting 1mbps. etc.
ascii_field: (on
the former, no way
to flip a bit - physically - at all; on
the latter - need high current and moderately high voltage)
ascii_field: both are immune
to anything short of room-melting level of radiation
ascii_field: (mask rom for known blocks from
the past, antifuse for future ones)
mircea_popescu: because we live in
the best possible world built on
top of !!!science!!! so nothing is ever sold by any sort of spec.
decimation: ascii_field:
the problem:
try finding reliable statistics on your consumer grade hard drives
ascii_field: best one can hope for is
to get it an order of magnitude below 'asteroid hits your computer' probability
decimation: in noisy channels
there are no guarantees
ascii_field: mircea_popescu:
that one's easy - infinity
mircea_popescu: there are many considerations here. for instance, as an exercise
to
the
trainee : calculate
the minimum storage space required
to GUARANTEE bit=wise identity over X bytes of information.
decimation: I do give a shit if none of
the pogos have a coherent record of
the blocks
decimation: or at least, in
the disk-storage gismo
decimation: actually,
this is an excellent case for putting ecc in bitcoin
ascii_field: decimation: i'm pretty certain
that i experienced a bitflip last night
decimation: you know with
these 'blockchain bitrates' I would be concerned about bitrot on disks
ascii_field: aha,
that's about right for 10min, 1MB blocks
mircea_popescu: anyway, it is interesting
to point out
that
the first 2 gb blockindes fit a whopping 188k blocks
jurov: Jautenim: problem was here, all
three came
through
decimation: mircea_popescu: but seemingly everyone is a "crypto enthusiast"
today.
Jautenim: gah, I didn't check
the new stator.sh
mircea_popescu: i am willing
to
take any odds on "he doesn't know
this exists"
decimation: ascii_field: one wonders why bezos didn't order
the creation of his own rsa code
mircea_popescu: i
think some implementations end up with different gaps between
them or something.
decimation: although it seems
there might be disagreement between nodes in later blkxxxx.dat files
mircea_popescu: decimation well, it's so
to speak "unstrained" if you're
there, and "strained" if you hear about it later.
decimation: in
theory, if orphans are removed, we should yeild
the same blockchain
turds
decimation: mircea_popescu: bottom line, satoshi stores a binary blob of
the blockchain AND also
the bdb database
mircea_popescu: i was doing like 5
things at once past coupla days and so ended up leaving all sorts of
things unexplained fully.
ascii_field: (thing stows blocks in real
time, never erases. see src)
ascii_field: mircea_popescu: iirc i explained
this last night
assbot: Logged on 01-07-2015 00:25:18; mod6: i have 3 full-sync'd chains from between January and March, all have
the same blk0001.dat hash: sha256sum blk0001.dat \ 7aac5826b91b4f87a2e9534e0e38e8d64ed21aff8a4eb8ff8dde4e726e67fe1a blk0001.dat
decimation: ascii_field: shit, you are right,
the amazon s2d is just a pretty
TLS wrapper around
the openssl crypto
turd
ascii_field wonders if it was
transmitting into a philippino buffer
ascii_field: mircea_popescu: your node continues
to emit bursts
ascii_field: and as someone pointed out in last night's
thread, let's compute
the energy lost by failing
to boil visa, mastercard execs into biodiesel
mircea_popescu: nah,
they had
the "unsustainable" crapola on
the back burner for a coupla years, as
they were hoping
they may actually elbow
themselves into relevancy all
through 2013-14
ascii_field: mircea_popescu: what is peculiar about
this piece? reads like
the same old crap