anyone seeing packet loss this morning?
anyone ever seen a packet? :)
har, har
any update on the kvr17 reboot?
pjs: currently thinking that may be hitting some bug that wasn't before.
@weather -v yyz
mnathani_: Verbose results will be PM'd to you.
@weather akl
Auckland International, New Zealand: Light Rain ☂ 55°F (13°C), Humidity: 94%, Wind: From the North at 9 MPH -- For more details including the forecast and almanac, see http://www.wunderground.com/cgi-bin/findweather/getForecast?query=-37.00805664,174.79167175 or re-request this with: @weather -v akl
ahh, yeah, it didn't feel too bad now
but damn that humidity
For 13C and rain? High humidity really is to be expected
@weather New York City, New York
New York, NY: Overcast ☁ 82°F (28°C), Humidity: 53%, Wind: From the WSW at 3.0 MPH Gusting to 11.0 MPH -- For more details including the forecast and almanac, see http://www.wunderground.com/cgi-bin/findweather/getForecast?query=40.746399,-73.981598 or re-request this with: @weather -v New York City, New York
Oh they're not bad today. I've seen NYC with 80%+ at temps around 30C, and that is just awwwful
it's not actually raining
it did a little earlier
yeah 30c + humidity has to suck
oh it looks like openbsd's early 5.8 release is about 20th year anniversary, and not some hackathon or such.
@weather
qbit: Not sure what you're looking for there, chief.
bah!
@weather rye, co
Rye, CO: Clear 87°F (30°C), Humidity: 20%, Wind: From the SE at 4.0 MPH Gusting to 13.0 MPH -- For more details including the forecast and almanac, see http://www.wunderground.com/cgi-bin/findweather/getForecast?query=37.983894,-104.865448 or re-request this with: @weather -v rye, co
so am i missing anything from having weechat 1.1 instead of 1.3?
anything significant
umm, dunno
http://weechat.org/news/89/20150816-Version-1.3/
you're missing that
up_the_irons: You available?
http://weechat.org/files/changelog/ChangeLog-1.3.html or that specifically, since it's a full changelog
yeah doesn't look major
RandalSchwartz, you notice that a lot lately? i used to use portaudit (until it all merged) but never got as many alerts
(pkg audit)
heh i was wondering what grody was talking about :)
some reason my buffer was still on a day ago :/
slowly catching up
yeah, i agree. didn't see any bugs that i've encountered (like the filter thing that happened twice)
where i couldn't untoggle or retoggle my filter depending on the channel i was in. restarting weechat didn't fix it. super weird.
haha
ok it's even easier to update now
i've updated :)
you used to have to run something on command line as well i thought
yeah i'm not even sure what the filters thing is about
i just use basic features
Filters are the whole reason I use Weechat! specifically, smartfilters
mercutio: there is an upgrade path wherein you save your state to file (/upgrade -quit) and then relaunch the new binary
For those platforms that don't support in-place upgrades
yeah you can jut type /upgrade
oh right
up_the_irons: any updates on that disk?
brycec: same here. otherwise, i probably would have just stuck with irssi
qbit: i can answer that for you, should be later today.
hopefully
It's been like 2 weeks now... Very disappointing.
what has?
Since qbit ordered the dedi in the first place.
oh wow
Took over a week to get the dedi, and then the disk thing...
Not up to ARP's usual standards, for sure.
(And qbit should know, he's been a longtime customer)
true story
yeah i'm not sure why provisioning took a while
damn raid resyncs take ages
even with 3tb drives they're slow, i suspect 5, 8 etc tb are going to be even worse.
brycec: we've had very bad luck with TWO batches of drives.  over 50% of what we ordered has failed.
qbit: toeshred is going to the data center now and will be replacing that drive for you.
up_the_irons: thanks
qbit: np
Ouch, that's sucky
At a previous job, after a bad batch like that, I rolled out extensive burnin testing.
Shouldn't "have" to do that but whatever
(There was also a lesson learned about thermal shock to hard drives around the same time)
protip: Don't power on a hard drive from 40F.
@calc 40f to c
is that how you do it?
convert 40 °F  (degrees Fahrenheit) to degrees Celsius;4.444 °C  (degrees Celsius);277.59 K  (kelvins);499.67 °R  (degrees Rankine);3.5556 °Ré  (degrees Réaumur);9.8333 °Rø  (degrees Rømer);4.434 °C above triple point temperature of water (273.2 K);4.444 °C above melting point of water ice (at standard pressure) (0 °C);4.44 °C above ice point of water (273.15 K);temperature;[temperature];Thermodynamic energy E from E = kT:,  ->24 meV
It was winter, the storage unit was outside, etc
hmm i would have thought a disk would be fine at that temp
brycec / up_the_irons at a previous job I had 110% of some gateway computers shit the bed over the span of a year
(Where winter = cold for us in the northern hemisphere)
yeah i can't say i've ever tried turning a hard-disk on at such low temperatures
good luck finding a data centre that low :)
mercutio: the shock of going from cold to warm in such a short span of time was *bad*
I mean, you could probably run drives at that temperature, provided you were /keeping/ them at that temp
worse than 20c to 40c?
but powering it on and letting it warm itself up...
i used to have a hard-drive that wouldn't work right until it had warmed up
brycec: yeah, mercutio and I have been talking about implementing a burn-in process
In short, yes. Google it.
learn something new every day :)
up_the_irons: there is an iso someone made for burnin / power testing
qbit: oh yeah??
http://www.stresslinux.org/sl/
qbit: hard-disk in particular?
or cpu etc?
full system iirc
cos i never seem to see cpu problems unless someone didn't attach heatsink properly.
it also may have been a iso I built.. hard to remember
but OpenSuSE, ewww
I made a burnin ISO for work, did simple mprime and smart self tests kinds of stuff
(not that it helps because I can't distribute it)
i like it how memtest is bundled with ubuntu etc these days
i tend to run that for a while
yeah, it's super easy to make them.. especially with deboostrap and friends
handy for power tests too
made one for oracle's facilities guy
so blame me for cloud pricing \o/
i saw that solaris is adopting pf
but mostly blame larry
is
i keep kind of hoping solaris will die
now that oracle control it
wow. cool. pf rocks
me too
there's opensolaris
and illumos
i thought all their main engineers left.
and smartos
but there must be some left :)
all the good people left
and that
it's like the skeleton crew for mysql
i hated sun a lot less than i hate oracle :)
yeah sun seemed decent
except for the way they got possessive about java
thus leading to the openjdk
yeah they weren't perfect
but bearable :)
even before they became snorkle
oracle is unbearable :)
they were kind of innovative too
i don't really see oracle as being innovative.
maybe before rdbms was a popular thing.. but now it's just a jerk org
apparently oracle doesn't have an autoincrement column type?
neither does postgres?
postgres does
foo serial
ohh
foo int serial
it creates the required incrementor, and the default value for that
yeah, just like the standard says.
because sqlite adopted that too
it's not part of base sql anyway.
whenever Richard Hipp wonders how to interpret the SQL standard, he just asks himself "what would Postgres do?"
and reads the postgres docs :)
so sqlite is really just a postgres clone
has serial always existed in progres?
I think so
i thought you had to create sequences.
you *can* create sequences
and then assign them as default values
but serial just does the right thing
ahh ok
so what do you do in oracle?
you do that explicutly apparently
lemme find the reference
mercutio: memtest has been bundled for ages.  I remember running it on our very first kvr host like 7 years ago.
up_the_irons: ahh ok
i only started using ubuntu around karmic
and i suspect that's about where kvr hosts started?
Hmm.  not finding it
it wasn't in debian though
ok serial type was added in 1998 to postgres
i wonder when debian got that version :)
ahh... https://metacpan.org/pod/Rose::DB::Object and look for "Oracle"
Since Oracle does not natively support a serial or auto-incremented column data type, an explicit sequence and trigger must be created to simulate the behavior.
that surprised me though. :)
I think 8.3 and up have serial
this source said 6.4
mercutio: i do dev crap on the identity management team
nice!
go pg
<3 pg
yeah i dunno i used postgres a bit earlyish
but then it seemed that everyone wanted to use mysql :/
even for just simple web hosting
protip: "Always use PostgreSQL" - Oracle Employee
cos it faster for reads or something
mysql was initially faster
but because it had less integrity
update/insert is what matters anyway
postgres really was ACID
can't you just use transactions?
ha!
transactions didn't work with the *fast* backend
just the *slow* backend "innodb"
oh right
so people deployed isam to get the wrong answer fast
hahaha
the thing is
lol
often with slow web sites it's php not the db that's slow
aside - someone actually implemented a bridge to use *postgres* as a *backend* for mysql
mysql makes me rage because of the truncate vs error thing
sinec the backend is pluggable
that's just cray imo
heh
I think they said it was faster than innodb :)
does anyone still use postgres for web sites?
many people
including me
and my clients
with wordpress etc i mean
old company does
custom stuff i can understand :)
in fact, if you're a greenstart, pg is really the only choice
all modern web frames can use pg
ah interesting
i remember a few issues with "mysql" code that didn't work on postres.
thanks to the modern php dbi layer
postgres
well custom code might not port
but if people are keeping it in mind i suppose it's not so bad
but the frameworks all got smart
yeh actually
frameworks have progressed
plugins might be the limiting thing on wordpress
but do people still create custom sql queries
or are there query builders
i've been too out of touch with that stuff
I use Rose::DB::Object and DBIx::Class in Perl
both of which are backend agnostic
Randal :)
it's the guy you hire to write your repetitive SQL
brycec: so hand-written code could mean you use mysql/postgres specific functionality.
haha
otherwise you spend hours writing the SAME DAMN THING with just different column names.
i must admit i'm not a huge fun of database coding
design etc
Maybe I do use X-specific functionality? I'm rarely concerned with being database-agnostic :P
but whenever i see tables with 30+ columns i'm like uhh
But then again, I'm not writing for millions of users. Or really even more than 1 user...
whenever I see a table where some columns end in digits, I run... very fast.
RandalSchwartz: like a year?
lol RandalSchwartz
Like "col0" "col1" "col2"
brycec: true, but i mean you can justify it, but many others are probably doing similar.
oh
i used to do that with non-sql :/
c, pascal etc.
sure
but in SQL, you must understand projection (join)
heh
yeah that's the idea behind sql
I have a great story about that, but I don't want to take the time to type it here
i imagine dbase3 with indexes would work just as well for most stuff though
because no-one uses the relational stuf
simple version: gas hauling... 1-n pickups for the tanker truck, 1-m deliveries
well, hardly anyone
guy who got stuck who I replaced was writing multple columns in each record for each pickup, and then the same for each delivery.
they fired him, brought me in
I wrote a simple join.
i think part of the problem is that it's hard to understand database design for a lot of people
and the few that understand it well aren't good at making it easy to understand for those that don't
hey, I resent that
haha
although I happen to be good at explaining complex things simply
do you teach db design?
I have
just brownbags
RandalSchwartz: well you might want to keep in mind that i'm kind of pessimistic about these things :)
I've taught Perl-DBI
i think it's php coders that need to learn it :)
i suppose it's kind of like designing larger programs
as scope rises, the ability to screw up greatly rises, with overdesign, and wanting to rearchitect the whole thing, but wanting existing code to ekep working
qbit: your drive has been replaced.
oh nice!
mercutio: thanks!
showing up ok?
uh.. gimme a sec :P
all good :)
mercutio: I assume you didn't power cycle it?
yeah
no power cycle
k, phew
installer is loading..
cool
seems ok so far
so it's showing fine
and working
did it screw up before install finished before?
RandalSchwartz: are you RandalSchwartz of perl fame?
mercutio: ya
io errors
yeah that's a very quick failure
was
second boot had no disk at all
mercutio: getting io errors on the first disk now :P
not even kidding :P
what kinda disks are these? need to not ever by them ever
Western Digital Re (Enterprise)
wd re4s
no, Re's, not RE4's
well they dropped the 4 on recent ones
but it appears to be the same
that's true, but they are the latest generation
up_the_irons: the model code is still the same on smart
mercutio: OK
but yeah hard to know if there's any diff
just different locations can make a diff in the past apparently
ok qbit will get the first disk swapped out
thanks.
should be shortly
qbit: these drives are generally good, never had issues until now.  I think it's either a bad run, or someone dropped them shipping (twice)
qbit: would you be able to paste the IO errors?
ya
sec
are you ok with first hard-disk being hard-removed?
yeah, haven't been able to build the box because of .. io errors :P
cool
http://i.imgur.com/TVbhilk.png
er
http://i.imgur.com/TVbhilx.png
yeah that looks like a failure
qbit: thakns
*thanks
yeah that doesn't look good
yeah
qbit: first disk is swapped out
woo
box, Y U NO BOOT CDROM!?
well, it still had grub on it.. not sure if there was some zfs magic that was making the mbr show up from the second disk.. or if the first disk ... had the same boot code on it...
usually grub is on both disks
at least with linux
magic
there's multiple ways to go
and arch linux describes in depth the bootup stuff if you're into that kind of thing
I have received some pretty terribly packed hard drives on the years
toeshred says these were packed pretty good though
i've always received pretty well shipped hard-drives
i think it's dropping g-force damage
err packaged
I'll see if I can find my picture of a hard drive that turned up loose in a large cardboard box with just some newspaper supporting it :(
gizmoguy: from where?
I can't remember
was a few years back now
ahh
The drives that were left in their original WD shipping (and shipped directly from WD) have been fine in my experience. But when you let Newegg or Amazon lackeys just toss drives in a box loose with a handful of packing peanuts, I fly off the handle.
gizmoguy++ I've been there
gizmoguy: pbtech? :)
brycec: yeah buying from computer specialised places probably makes more sense
brycec: that's why I like to buy the 20-pack drive bundles.  They seem to be sent in the 20-pack bundle straight from the manufacturer in that fancy shipping tray.
up_the_irons++ (except even still sometimes...)
i find drives tend to just come in plastic tray things with a anti-static bag over eeach individual drive
I have a half dozen of those trays hanging around myself
handy for unloading + moving computers
yup, I keep them too
now i'm curious
if i take an old hard-disk with a few errors on it
how many times, and from how high can i drop it
These came from Amazon http://imgur.com/RecEs0R,SgkdIXQ
mercutio: I've tested this :D It doesn't take many drop-kicks... usually 1.
I *would* have bought a 20-pack of the WD Re's, but they are sold out everywhere... (only individual drives are available)
brycec: that's pretty fucked up
For those unfamiliar with the packaging we're talking about http://epeusa.com/images/HDD-Bulk-Pack-WEB.jpg
up_the_irons: Yup. I didn't even waste my time, they went straight back.
lol
that's awesome
one problem with those plastic trays is they don't protect from two sides of a box that well.
brycec: oh you have? :)
they protect from side to side movement, but not really if the drive lands on the head or foot.
mercutio: I took out my frustrations when working tech support...
Put some dents in the office walls by accident :)
heh.
oh i've seen smaller ones too.
so far no errors
well that's good :)
hopefully it stays that way
mercutio: smaller stuff like http://thumbs1.ebaystatic.com/d/l225/m/ml3FTH7nzz5t9vKqYEM-q8w.jpg you mean?
yeah
these images are so tiny
i'm thinking clera plastic
somewhat translucent
but my google foo is failing me
oh even ebay loads small
http://www.servethehome.com/wp-content/uploads/2011/05/Newegg-Single-Hard-Drive-Packaging-May-2011-Proper-Inner-Box.jpg
i've seen that method before too
bah can't find it
up_the_irons: any chance I can get my bill to reflect now as being when I received my service? :P
I think he just left for somewhere a few minutes ago.
k
how are the new drives holding up?
so far good!
qbit: yes, certainly
thanks <3
@weather yyz
Toronto-Pearson International, Ontario: Mostly Cloudy ☁ 68°F (20°C), Humidity: 56%, Wind: From the SW at 12 MPH Gusting to 18 MPH -- For more details including the forecast and almanac, see http://www.wunderground.com/cgi-bin/findweather/getForecast?query=43.67722321,-79.63055420 or re-request this with: @weather -v yyz
@weather hell
Hell, MI: Overcast ☁ 59°F (15°C), Humidity: 74%, Wind: From the West at 6 MPH Gusting to 10 MPH -- For more details including the forecast and almanac, see http://www.wunderground.com/cgi-bin/findweather/getForecast?query=, or re-request this with: @weather -v hell
@weather smo
There is 1 weather alert in effect for your area! There is a Special Statement.
Santa Monica Municipal, CA: Clear 72°F (22°C), Humidity: 78%, Wind: From the SW at 7 MPH -- For more details including the forecast and almanac, see http://www.wunderground.com/cgi-bin/findweather/getForecast?query=34.02099991,-118.44708252 or re-request this with: @weather -v smo
Special Statement!
"You. Have. No. Water.  G'bye."
I believe the answer is "Yes" 18:25:16           qbit | RandalSchwartz: are you RandalSchwartz of perl fame?
yes
sorry didn't see that question earlier
But RandalSchwartz, your adoring public awaits!
they are used to this. :)
qbit?
For whoever's interested, I've got h2o running on a development Mac, and it was a far easier process than nginx [for how I'm running the web server, as a standard user]
nginx has a hard-coded set of paths it expects permissions to unfortunately
@google h2o webserver
Searching for 'h2o web server' instead.
83,300 total results returned for 'h2o web server', here's 3
h2o/h2o · GitHub (https://github.com/h2o/h2o) Subversion checkout URL. You can clone with. HTTPS. or. Subversion .  Download ZIP. H2O - the optimized HTTP/1, HTTP/2 server. https://h2o.examp1e. net.
H2O (https://h2o.examp1e.net/) H2O is a new generation HTTP server providing quicker response to users when  compared to older generation of web servers. Written in C, can also be used as ...
H2O the HTTP/2 Web Server @ Calomel.org (https://calomel.org/h2o.html) Jul 24, 2015 ... H2O is a fast and secure HTTP/2 server written in C by Kazuho Oku. ... This  configuration is a stand alone. static file web server listening for for ...
yeah it's showing a surprising amount of promise
i like how they're trying to get push caching etc done well
ie they're not just trying to serve 10,000 copies of some 10k static document as fast as possible
there's some huge complications with push caching though
like how does the server know if you need a copy or not?
one idea is you use a cookie to guess
i wonder if something like the icp protocol that squid etc uses is good
basically hash documents and you'd send the hash of what you have
I'm not familiar with h2o's internal design.  I'm using it as a reverse proxy to io.js apps.  What do you mean by "they're trying to get push caching etc done well" mercutio?
well http2 allows servers to push data, and prefill the cache of web browsers
so that the browser can then request it and get a cache hit
but the problem is that some of those requests may already be in the cache, and so you may end up sending all of the data when normally the client would just request any different version with if-modified-since.
Is it correct to assume HTTP2 features are only enabled if the config file contains references to them?  I've only just now got h2o running on port 80
It's so much more straightforward to set this kind of stuff up on FreeBSD
you need ssl
then it'll just work
it may work with http too, but no browsers do http2 on http
I'm not interested in HTTP2 ATM was my point
I like what I see so far though
@exch 1 nzd to usd
1 NZD -> 0.64922586308086 USD (as of Mon, 24 Aug 2015 23:00:09 -0700)
apparently stock markets are a little crazy atm