some day, I will meet the wise and noble up_the_irons
he should just drop by Maria's Downtown, and that'd make it easy. :)
RandalSchwartz: where is Maria's?  I'm in downtown so little now, having specifically engineered all my gear to be remotely accessible (if something is down, there is no driving to add to additional downtime)
I'd say I can count that as a success :)
RandalSchwartz: are you still in LA?
all: today I will send all the pre-order people their signup link
I arrived the 4th, leaving the 16th or so
maria's is 600 block of floswer
flower
ah cool
oooh, and being mayor on 4square here gives me a free appetizer and dessert!
yeay
LOL
but since I get double miles for my credit card mon-wed, I'll save those for thursday :)
I've got neil's new machine mostly configured... he just has to test and approve before we go live this weekend
and then we can get off that ugly expensive EC2 box
ah cool
how is EC2 pricing these days?  I find it kinda confusing so never did real calcs...  but from what i've heard isn't it comparable to a vps, like $20 / month for something basic?
like if you add up all the "CPU Hours" and crap
for a full month
sup
up_the_irons: I think a basic EC2 instance costs around $70/month.
Ask cmeiklejohn for the details. He's done the calcs.
neil's paying about $120/month for about half of what he'll be getting from ARP
so it's about 2 or 3 to 1
and that doesn't count the multiple redundancies
what are the multiple redundancies?
mhoran: roger
power supply, disk, network, ultimately the CPU,
EC2 doesn't promise restarting my instance if the physical machine dies
I suspect you would
RandalSchwartz: doesn't EC2 have redudancy in those things though?  I mean, it's supposed to be "cloudy" ;)
no - they promise restarting the boot image
RandalSchwartz: well yeah, I'd copy all the instances to a new machine
but not the live disk
RandalSchwartz: oh i c
we've had to calculate that into backup plans
so the live disk is not persistent, in that case?
right
weird
I have a script that snapshots to permanent storage
i mean, i assume they still have the live disk stored somewhere
but it'd still be a pain to get restarted
Yeah - weird eh?
you can't boot from permanent storage
only attache it
ah ok
I thought you could by now
nope
so my / is always volatile
I can issue a reboot, and it's preserved
but if they need to migrate, it's dead
that seems really dumb; but i think the EC2 market is different than the 24/7/365 uptime machine market
yup
it's all about servers starting from known AMIs
gotcha
so / is really irrelevant
yeah
what's important is where your DB is
and what your IPs are
gotcha
in fact, you coudl view local changes to / as distracting
and possibly destructive
you're supposed to get everything as you want for a boot image, then make an AMI from it
and then launch everything from there
it's a different model from a VPS
RandalSchwartz: ah ok, so you can make your own AMI's, but not have an "forever evolving" AMI which would be like a VPS, always at the last state you had it
right
you *can* reboot your own server
but they don't promise eternal existance of it
so it's possible you'll have to boot from an AMI
as I said, you offer the far better deal for most mortals
and I really appreciate getting off the single dedicated box in texas
because god-forbit that thing go down... it'd be my problem to have backed up to another disk somewhere off box
where I know you have my disk in a place that a new CPU can look at it
and since I'm not the only guy on the box, many people would be yelling at you :)
so I don't have to be the guy yelling
RandalSchwartz: haha, yeah that's true
a buddy of mine had his dedicated box die recently; he then just bought a vps with me
there's an advantage in numbers
yeah
I know there's a slight security risk in the ring0 being penetrated, but that's all worth the upside
yeah
i fear that enough to take measures to avoid it
have you seen any recent hacks that break ring0
theoretical ones have been demonstrated
yeah, just wondering about things in the field
you said you'd be rebooting when you needed to update.  I don't recall any of those.
the debian / ubuntu security mailing list shows them from time to time, but when I see "[so and so] has demonstrated [such and such]", i never know if that "hack" was disclosed or still secret
I tend not to want to reboot every time a security update comes out.  Many are not applicable.  The ones that are can build up into a "super" update.  Unfortunately, if I rebooted every time a Linux kernel hack came out, it'd be every few weeks
yeah - that's what I expected.
I guess you don't do that
the break-ins we've seen in the past (like the fsckvps fiasco) tend to come from weaker links, like bad / common passwords, leaving this too open, etc...
ok
yeah, it'll always be low-hanging fruit
s/this/things/
hence, no more sshd on 22 for me
yeah me either
neil's first box was getting hammered 24x7
so I just removed that
dictionary attacks, coordinated from 30 IPs
so clearly it was a mass attack
yeah
a .. z but from rotating 30 IPs
bad guys got too much resourcs
I blame redmond
same
for all the good that having windows made business buy computers...
I wonder if enough bad has already been done to reset that
gotta wonder
considering that 80%+ of spam is from zombies...
and zombies couldn't exist without windows desktop idiots visiting pr0n sites
gotta return some movies to the video store...
whats up
up_the_irons: you doing signups at 12am? :)
ok work now, gotta run
lol
that would be sweet