133400+ entries in 0.069s
mircea_popescu: key
to
the split is a rewrite of
the db so it's less fucktarded.
mircea_popescu: there's absolutely no conceivable reason
to have all
these 3 items in
the same place.
mircea_popescu: anyway, continuing
the
trinque discussion, it seems entirely unavoidable
that
trb will become 3
things : a wallet node, optimized for pumping out local signed
tx ; a block node, optimized for keeping
the blockchain, getting blocks, no mempool nonsense ; and a spy node, optimized
to keeping
track of
the lies and nonsense flowing
through
the relay network (mempool,
timing nodes, what have you).
☟︎☟︎☟︎ freetlas: In case you're
talking about people from Venezuela
BingoBoingo: Anyways, accomodating
the
truth
that most people want a friend more
than I do, and letting
them shoulder
their share of
the burden helps make BingoBoingo spanish functionally amusing in person.
BingoBoingo: Because
the venezualana is
too busy
to deliver suffient humilation on her own.
mircea_popescu: lmao watching BingoBoingo spanish is amusing. los republica ? how many of it are
there!
BingoBoingo: <asciilifeform> all i can say is,
to date only observed db death on boxes with actual bit rot. << Aha, saw when
trying
to start
trb from blockchain loaded onto box from corrupted
thumb drive
a111: Logged on 2017-12-21 16:44 mircea_popescu: in other sads : one
trb node was dead since fucking 22nd of august, because -- it ran into
the fabled "terminate called after
throwing an instance of 'DbRunRecoveryException' what(): DbEnv::txn_checkpoint: DB_RUNRECOVERY: Fatal error, run database recovery" which
then kicked in a script
to clean it up, which it did, but couldn't boot back up because for yet-unknown reasons
there was a spurious .lock leftover ; corner case u
BingoBoingo: with an under 512 kbps cap, not many other nodes will want
to be friends
BingoBoingo: 1-1.5 mbps cap
tends
to allow resonable block reception
time and support a number of connections, and keeps pipe faily clear outside of block
time
mircea_popescu: in practice more like 15x
that, because it gotta chatter like a boring housewife.
mircea_popescu: ie, unless you got a good reason running a centralizing node, capping
the bw for
the damned
thing doesn't do anything observable.
mircea_popescu: and
the narrow sliver in question is in
the kbps range
mircea_popescu: bb broadfly has it,
there's a narrow sliver of bw
trb/prb can even use ; more
than
that it generally wastes.
BingoBoingo: Once
trb gets enough network speed
to work, extra doesn't offer much
mircea_popescu: by
the
time your
trb node eats 10 MBps even, you've more serious problems
than "buy more pipe". and
that's inst not sustained.
BingoBoingo can't recall running a
trbtron I could
touch on a connection faster
than 3 mbps until in Montevideo
mircea_popescu: asciilifeform i don't see a number in
there ; be moar useful and speak
the value.
mircea_popescu: asciilifeform he's getting his foot in
tyhe door ; it'll scale.
BingoBoingo: asciilifeform:
The quote is for 100 MBPS decicated symmetrical. If
there's more
traffic in one direction, asymetrical dedicated lines are also available.
BingoBoingo: trinque: As far as
the racks
themselves. Ready within 72 hours of signing a contract and paying for
them, but IP address block from LACNIC will
take 2 weeks from application
to allocation.
mircea_popescu: ncovered, 1k man-hours worth of spuriously idiotic automation down
teh drain.
mircea_popescu: in other sads : one
trb node was dead since fucking 22nd of august, because -- it ran into
the fabled "terminate called after
throwing an instance of 'DbRunRecoveryException' what(): DbEnv::txn_checkpoint: DB_RUNRECOVERY: Fatal error, run database recovery" which
then kicked in a script
to clean it up, which it did, but couldn't boot back up because for yet-unknown reasons
there was a spurious .lock leftover ; corner case u
☟︎ BingoBoingo: trinque: More slowly
than I would like.
The whole US citizen bullshit is giving
the accountant quite a workout when it comes
to making banking work
to pay for
the racks.
BingoBoingo: Where's
that
trilema about cock eating rock
trinque suddenly realises
that
there are
two intuitive items,
tit.. and cock
trinque: asciilifeform: I don't blame
the patch, only "now I have an observation which suggests
touch
the knob"
mircea_popescu: asciilifeform meaningless magic number got set
to
the "closest ot inexistence" available.
trinque: indeed, should be
tested independently of
this, can possibly preserve
the high version.
trinque: should be
trivial
to add an rpc
to change
the var
trinque: I agree with
that as well.
a111: Logged on 2017-12-21 05:34
trinque will be comparing multiple
test subjects.
trinque: asciilifeform: where was
that in argument
trinque: I have
the republican list in addnode and
that's it.
trinque: mats: my node.deedbot.org node *immediately* synced
to
top after weeks of restarts and praying
to satan when version number was set.
trinque: in
the case of
the version number (which I will obviously
test extensively, and said so) it's possible it's a bad move
to crank
to maxver, because it's a lie saying "we support all misc novelties below
this int"
mats: in regular practice i'm uncertain how
to masquerade as prb without running a heathen node and plugging
trb into
that
mircea_popescu: somehow nike manages
to get its shoes sewn ; but not us
trinque: I reported
this in
the log
trinque: this sort of
thinking creates gigantic blind spots in which ways 0.5.4 is broken will hide.
mircea_popescu: asciilifeform in his case, because he wants
to at least occasionally get
txn out
trinque: think for a sec,
that with
the high version number you're advertising support for protocol "features" you
then immediately issue a ban for.
trinque: one which probably doesn't implement enough of
the prb protocol
to get banned
mats: and at
that
time i just fed
those blocks back
to
the advertised nodes, it caught up
mats: i've had some luck syncing with 'btcwire' nodes, when all of
the advertised nodes were ~30 blocks behind or unreachable on sunday
trinque: I was brief on details because I fully expected
that others are seeing
the same behavior with
the
trb nodes
they maintain. public indexes of nodes report
that just about every single one is far behind.
mircea_popescu: for instance, what specific variable was
to be measured, and how.
mircea_popescu: trinque i gathered you moved from knob set
to x
to knob set
to y ;
that's not
the same as a "do
this experiment".
mircea_popescu: "study
this room" rarely works in my experience (and yes, i experienced ; we have small
toys for hiding etc)
trinque: I plainly said what I was going
to do about it
trinque: what was meant is
that I want others
to study
their nodes while I study mine
mircea_popescu: trinque
the problem with "mats also reported
to me
that his nodes *will not keep up*, ben_vulpes also" is
the ambiguity attendant.
two guys walk into a diner, agree
that coffee sucks
there -- one's a colombian, local brew is swill ;
the other's from alaska, can not stand hot beverages.
BingoBoingo: Well, I just stayed a playful friendly extranjero and let
the laughing and
the context do
the not heavy at all lifting
mircea_popescu: you got a blog, you saw
teh fetlife/okcupid/whatever write-ups, knock yourself out in a mead of fermented detail.
a111: Logged on 2017-12-21 05:23
trinque: for now, I can report
that setting
the version number
to 50400, a
trb node will catch back up,
txns will unstick.
trinque: I already put it in
the log, what