201300+ entries in 0.12s
mircea_popescu: asciilifeform possibly
those were just fake user agents ?
Framedragger: in which case, may not want
to do
that
till caching is in place, or sth. main
thing is
to have stats be accessible which is now cached anyway
mircea_popescu: you (all of you, really) are very well linked since all
the
trilema mentions and etc. we've built like a coven
mircea_popescu: google will definitely slurp 5mn pages in
the next week-ish
Framedragger: asciilifeform: i also wonder if phuctor site my come under additional stress now
that you removed
those 'Disallow' entries from robots.txt :/
Framedragger: pete_dushenski: plox
to update bot commands at some point,
ty! :)
mircea_popescu: this is also
the first
time a mysql db crashed in my personal experience.
mircea_popescu: my first
time using any of
these shits. prior
to
this morning i had nfi what /var/lib/mysql/ is or how you repair crashed mysql db or anything of
the sort.
Framedragger: well
the default for DBs is
to store
things under /var. /var is under root unless specifically separate partition; or you
tell db
to store under /home/mircea/dbdata or sth.
mircea_popescu: i still have nfi elephantiazis hit
that poor file
that it ended up 40g, but anyway, im moving
the whole shebang on a
tb partition and will see wtf myisamchk does.
mircea_popescu: Framedragger for some reason root gets 50G and mysql for everything is stored in
there.
Framedragger: mebbe you had a separate /var partition from linux install long ago, mircea_popescu..
this can bite
shinohai: "Enphase's storage
technology is optimal for
the masses." <<< Indeed!
mircea_popescu: anyway,
turns out
the myslq files are stored on a
tiny partition for some incomprehensible reason. fancy
that wonder.
Framedragger: (mk /stats now updated every 30min, will see how
that one goes.
http://siphnos.mkj.lt/ updated
to include link
to
the cached copy. will see if
this gets indexed by google (main index page already is)) :)
BingoBoingo: <mircea_popescu> incredible.
this is even legal ? << O hai I R
the packet interceptor. Pls
to pay again for delivery.
jhvh1: hanbot: Bitfinex BTCUSD
ticker | Best bid: 1174.7, Best ask: 1174.8, Bid-ask spread: 0.10000, Last
trade: 1174.7, 24 hour volume: 29477.80773128, 24 hour low: 1120.3, 24 hour high: 1179.3, 24 hour vwap: None
phf: i usually pipe du
through sort or awk >n. i kind of wish
there was an "h" command instead
that would
take a column number an humanize
that column
mircea_popescu: 39491524 /var/lib/mysql/trilema_blog/tril_posts.MYD << reported by du | 40439263232 Apr 6 11:12
tril_posts.MYD << reported by ls -l
mircea_popescu: last i saw MySQL Disk Space 596.29 MB, so evidently mysql "repair
table" corrupted
the
table
to shit.
mircea_popescu: meh. in which we find
trilema's posts data is 40g somehow.
phf: it's pretty
tmsr
too. "if it can't fly, it doesn't deserve
to live anyway"
Framedragger: i'm waiting for
the day
that freenode goes down while datacenters with logotrons are experiencing issues. everyone's gonna freak out,
there will be a new more urgent push for gossipd, it's gonna be great!
mircea_popescu: 0 indication of what
this is. will it
take another 6 weeks ? "maybe". how do i
tell ? what do i do ?
mircea_popescu: load average: 0.01, 0.05, 0.21 <<< meanwhile i am... waiting... for it
to... repair.
Framedragger: in mysql it's called 'binary log'. dunno if on by default
tho
Framedragger: maybe it's going
thru full mysql
tx log if you have one
mircea_popescu: "every
thing you do will resolve 99% of
the problem at
the most". always a little bit left. gah.
mircea_popescu: entirely not fucking iron-bound problem in
trilema case for instance. wtf can "repair
table" need >1200 s !
a111: Logged on 2017-02-19 03:54 asciilifeform: (iirc we had a
thread where i described how corporate ameritards, if given a problem like phuctor, would happily soak up a few $mil and megawatt of iron)
lobbes: Or rather, my process
too slow heh
lobbes: mircea_popescu: Nah nothing really. I noticed you pasted
that link earlier but bot didn't quote. I
think you were just
too fast for it
Framedragger: asciilifeform: phuctor.nosuchlabs.com
timing out, do you
think
this could be related
to
the cron cacher? (GET /stats every 5min, GET /phuctored every 60min)
lobbes: Mp must've been faster
than it inserted into database
lobbesbot: Logged on 2017-04-06 00:05:55: <mircea_popescu> until
trinque's deedbot payment is on i dunno what lightweight
to recommend you
mircea_popescu: i can't comprehend what'd possibly
take
this long
to repair it.
Framedragger: (aok. oh well
then, light cigar while making DB destruction plans)
mircea_popescu: weirdly enough, also not
timing out. how
the fuck long can a
thing waiot for a lock ?
Framedragger: mircea_popescu: sorry i forgot or didn't know
that you actually in fact restarted it.
mircea_popescu: Framedragger you
trying
to
tell me
that process'd have somehow survived a mysql restart ?
Framedragger: 150? no just
the one which is still
trying
to import large blob (if still running)
mircea_popescu: so far
the
table is crashed, and
the repair
thing seems
to
take forever.
Framedragger: (then `FLUSH
TABLES` should remove any outstanding locks if still any; but don't
take my word for it)
mircea_popescu: why'd i want
to kill 150 processes by hand. let
them
timeout.
Framedragger: mircea_popescu: well you can still kill
the particular mysql process which is handling
the query, `show processlist` (may show "Waiting for
table metadata lock" under "State" column; "Info" column should display
the query in question),
then `Kill IDGOESHERE`
mircea_popescu: but
the box doesn't deliver,
theoretically, anything but a
text line now
Framedragger: there are ways
to list/remove locks but i reeeeally dunno if it's a good idea
mircea_popescu: no it's fine,
trying
to figure out how
to rescue
the
table.
mircea_popescu: ERROR 2002 (HY000): Can't connect
to local MySQL server
through socket '/var/lib/mysql/mysql.sock' (2)
a111: Logged on 2016-12-30 17:16 mircea_popescu: it really should be up
to operator wtf, if i want
to read dirty let me read dirty what sort of decision is
this for designer
to make.
mircea_popescu: turns out
this whole "read while dirty"
thing is "implemented" except notrly.
mircea_popescu: on
the basis of
too many dorks hammering it on
the www!
trinque: mysql's kinda notorious for
this shit
Framedragger: i wonder if it's due
to data
type misuse, or just mysqlol