up_the_irons: do you still honor the "EVIL" code?
jlgaddis: sure
probably gonna write an article over the weekend about my stats yesterday (72,567 total page views, normal average is ~2,000) and how well my vps/nginx held up.  =)
so if nothing else maybe you'll get a new customer or two out of it.  seems like you usually do whenever i write something about arp.
really fucking impressed w/ nginx.  several months ago, i decided to try it out when apache shit itself at waaaaaay lower levels.
jlgaddis: awesome, tnx!
yeah, nginx is pretty fast
Oop, you know what the music means.
jlgaddis is it static html or serving something behind nginx?
it's wordpress, so php
i use a caching plugin though
the first time $page is requested, it generates .html and .html.gz versions of the page. on subsequent requests for $page, it'll serve those up instead, as long as the cached version hasn't expired
also, all the static stuff (images, .css, .js, etc) is served up by amazon cloudfront
gotcha
wp-supercache or whatever
don't ask for an android charger at LA RubyConf
i got weird looks
then i asked for USB Type B cable (the regular standard charging cable)
weird looks again
I really, really need to find a new job. There are currently 5 people in a remote office working on the problem of "Is a machine powered on" for me :/
LOL
jdoe: what do u do?
jdoe, hilarious.
cue "how many admins does it take to find a power switch" joke
up_the_irons: I'm a sysadmin.
up_the_irons: they're ruby devs, what do you expect besides blank stares?  :P
up_the_irons: the remote office is in buttfuck nowhere, has no useful technical staff, and has poor infrastructure (flaky power/internet) -- so naturally I've been tasked with replacing their POTS system with voip.
kill me now.
jdoe, I am sending you a package which you should open right away. do not pay any attention to the ticking.
problem solved.
root cause: several of my coworkers can't distinguish betwen a 1U that's on and a 1U that's off.
taking the rest of the day off with a bottle of JD.
jlgaddis: LOL
jdoe: hah
do i need to bring out my x serve pic again?
sure. What is it?
looking.. my netbook is slow...
my only (recent) server room wtf is http://falseprophet.ca/~jdoe/fusing.jpg ... not pictured: the door that opens (literally into) that rickety table, that it was just sitting there for a long time, etc. etc.
can't find the "THESE WALLS ARE ASBESTOS, LEAVE THEM ALONE signs, next to giant holes knocked in the walls" picture :/
http://www.flickr.com/photos/51184165@N00/2174345059/in/photostream
omg
jdoe: what the F is that thing??
dunno what the proper name for it is, I've always heard them called fusing cabinets.
I believe in that case it's linking local stuff to CA*net (internet2 if you prefer)
jdoe: crazy
I just opened up 3 slots for 4GB beefy VPS', thinking of running a special at $89, or even $79 each; only three left, first to get 'em gets 'em!
i wonder how much 8GB sticks of FB-DIMMs are...
not THAT bad.
if anyone cares, I have almost 900GB of FB-DIMMs in my whole cluster, all Corsair.  Only 1 stick has failed.  So, go Corsair!
haha.
LOL
we've had a *lot* of that go bad.
jdoe: o'rly?
yeah. I think I've replaced at least one stick from every machine.
though it's not quite as bad as whatever's in the HP blade server.
oh wait, excuse me, not Corsair.  Crucial!
i always get that mixed up
I do like crucial.
been quite reliable
I dunno, I've never had problems with desktop ram the way I do with whatever's in these boxes.
haha
in my entire life,  I've ordered a shitton of ram for myself and customers.  I've found "crucial" to be good RAM.  and in that time,  only 1 stick needed to be returned.  and even better,   their RMA process was painless.  they overnighted the cross-shipment.
wow nice
yea good point
quality ram matters
i wonder what arpnetworks buys
ninor: up_the_irons just said they buy crucial
up_the_irons == garry == owner
oh
i used to use corsair
was really nice. but that was years ago
silicon mechanics? i used to use their servers too
they're only a few miles from where my datacenter was
er sorry, they're up here in Seattle
Yeah I've had good luck with Crucial too.
Had Corsair as well. No issues.
I really haven't had any RAM issues. So I'm a bad sample.
centos :|
ninor: I have nothing against SM. I suspect our woes are environmental and/or bad luck.
jdoe, oh i didn't even read anything. did you have a bad experience with some gear from there?
mhoran, would we even KNOW if we got bad ram?
i mean i wonder, how fragile is our data? speaking generally
there's bit rot on disk/disc, and in ram if shit goes sideways, how does that manifest?
depends on the level of "bad"  but you'd see data corruption,  odd behaviors, lockups, and possibly even "random" reboots.
you very probably wouldn't see any of that since it's all ECC ram...
i wonder if there's any way to put safety checks in operational data in the way zfs does for persisted data
yea but i mean, if you were to put checking for 'this' into software, what would that look like?
unless a stick dies, the only person who's likely to know is up_the_irons, and only because he's going to get log spam.
jdoe: *if* it's ECC ram,   correct.  but not all ram is ECC ram :P
no, but basically everything in a server you spent more than $20 on is.
im not even talking about arp
i'm speaking generally, in computing
oh oh,  in the context of ARP, and our vps's?  No..  we'd never probably know.
k i must be doing something wrong here.
right.
ninor: *generally* if it's non-ECC,  you'd likely see what I mentioned above.
k well sorry
YOU SHOULD BE >:|
haha.
k and what would a software layer do to manage that
and overlay it to build in resilience
nothing.
because you can't trust the software layer.
*sigh*
since it's running out of ram :P
software would have a hard time dealing with faulty RAM in any sane way.
lol
wasn't expecting that.
seems so.
ahh well.   /me goes back to editing the maps at waze.com