550000+ entries in 0.371s
mod6: but,
this is 10% of
the memory i'm used
to runing with
mircea_popescu: i
think it's more in
the 230k ish range
that
there's some large reorgs
mats: hmm...added
to my
toy wish list
mircea_popescu: still have
to unpucker
the cock enough for as much as a dribble
to come out
mircea_popescu: ben_vulpes:dealbook.nytimes.com/2015/01/22/winklevoss-twins-aim << yeah yeah,
they've been aiming at
the
toilet for
three years now.
mircea_popescu: asciilifeform: jurov, ben_vulpes, mod6, mircea_popescu - if one of you presently has a 0.5.3.x in a vm,
try running under hard constraint of 128 or 256m. see when croaks. << it doesn't croak if it has
the blockchain prebuilt,
that part's for sure.
mod6: basically; patches include so far, your patches, ben's rm UPNP &
the bdb config update patches.
mod6: asciilifeform: irc stuff is still in
there.
mod6: here's a script
that will build everything without
the rm_checkpoints patch, with exception
to
the version number update:
mod6: and so should you. I'll be
tabling
that, updating
the "Patch guide" and pulling
the
tarball with
that patch down from
the website as soon as my other vm hits current block.
mod6: in my most recent regression
tests (
this past week, i've been leaving out
the rm_checkpoints patch )
mod6: cool. in
this case, im running v0.5.3[base]+patches{1,rm_rf_upnp,2,3,4,6(db_config),unreleased-version-update-patch}
mod6: it would be neat-o
to see
this run on some sort of commercial router with a large flash-rom in
there.
ben_vulpes: i'm not entirely up
to date on
the
thread, asciilifeform
mod6: up
to 68,xxx blocks....
mod6: i'll put it on
the list here...
jurov: amazon sells
them between $20-$60 depending on version
jurov: rumor has it
that it has even slower i/o
jurov: it's faster
than rpi?
gribble: namworld was last seen in #bitcoin-assets 5 weeks, 5 days, 5 hours, 42 minutes, and 15 seconds ago: <Namworld>
That's 888poker
jurov: btw what about odroid? or it's
turd?
Apocalyptic: asciilifeform, nothing in particular, just want
to play with a board having a powerful CPU
Apocalyptic: alright, from what I got so far
the ODROID-U2 looks decent, with a Cortex-A9
though
Apocalyptic: asciilifeform, out of
these
thousands boards
to chose from, know any
that features a Cortex-A57 CPU ?
assbot: Why "representative democracy" doesn't work and doesn't make sense pe
Trilema - Un blog de Mircea Popescu. ... (
http://bit.ly/1Bj38sL )
PeterL: oh,
that was me restarting it again, be back up in a sec
mircea_popescu: asciilifeform how hard is
to field-upgrade
these rams ?
mats: jurov:it's more about acquiring useful muscle memory bits << kung fu is not a system of fighting, its a system of
thought. you must outthink your opponent, whatever form he
takes... because some of
them will be more
than just men.
mod6: awesome pics
today :)
PeterL: well,
thank me once it works ;)
mod6: well, i could be wrong. i just saw a
thing where
they were
telling some guy running a way forward version
that he should add more swap or some such nonsense. but yeah... so it's been in
there for a while.
PeterL: pete_dushenski I
tried
to put in a fix
to scoopbot, he should respond
to your blog now, let me know
the next
time you post something
infobel: any
thoughts on
Trezor anybody?
pete_dushenski: mircea_popescu your article's
title had me hoping for a deeper explanation into 'oficialismo' ...
kakobrekla: mircea_popescu i cant run .7 for more
than a day on system x, runs .8 endlessly
mod6: yeah, i've read on place
that
they have
this leakyness -- crashing problem all
the way up
to .9
mod6: a few days ago, I had been looking mapOrphanBlocks as a possible reason
that it runs itself out of ram as
this structure goes unchecked in size in v0.5.3 -- when I was at ~180,000 blocks,
these are
the stats
that I had built up so far:
mats: re: router discussion, seems
to me
the obv solution is an OpenWrt 'package'
nubbins`: kakobrekla let's see
the diffs :D
kakobrekla: mircea_popescu> did jack. < no leak in newer version, so
they actually fixed it
nubbins`: your handy pocket node is already
there
nubbins`: then eventually when
the real bitcoin core fixes
the current memory leaks
nubbins`: fuck, kill and restart bitcoind on a
timer
undata: pile up on itself when
the previous cron hadn't finished munging graphs
nubbins`: so maybe not design
the device around
the memory leak
undata: hah, I had munin on a micro and it'd help
take down
the instance
nubbins`: the inexorable consumption
thereof is
the issue
nubbins`: the amount of ram isn't really
the issue
mod6: ive got a separate vm
that has 1Gb of ram and it runs for about a 18-20 hours before it runs out of ram.
mike_c: i want
to say namworld, but not sure