sroute: cool, s/ip. pubkey sent...
-: sroute is wondering if Canadian back bacon doughnuts will every be available here...
***: Rada has joined #arpnetworks
nuke has quit IRC (Nick collision from services.)
nuke^ has joined #arpnetworks
mike-burns has joined #arpnetworks
vtoms has joined #arpnetworks
vtoms has quit IRC ("Leaving.")
mhoran: Guess that key wasn't in there. :)
up_the_irons: Ever work with any Juniper equipment?
Also, what are you using to route IPv4?
***: heavysixer has joined #arpnetworks
vtoms has joined #arpnetworks
vtoms has quit IRC (Remote closed the connection)
vtoms has joined #arpnetworks
heavysixer has quit IRC ()
heavysixer has joined #arpnetworks
jeev: i think ihad swine flu
and now it's gone
well i had something
mhoran: That's exciting.
jeev: very
-: sroute is still recovering from H1N1
sroute: eldest son also got it, he's more or less well now. Four days off school.
jeev: what were your symptoms
sroute: sore throught, headache, aches, congestion in upper chest / a little gurgling inside / fever for him, not for me, some cough. Comes on fast.
jeev: i wish i could get rid of sore throats instantly
sroute: s/throught/throat/
Feelin' tired; barely got any work done all week... and on that note I'm going to go make some tea
vxp: dongs
up_the_irons: mhoran: have not worked with Juniper, except drool over them. IPv4 is routed by a Cisco 4500 w/ SUPIV
mhoran: up_the_irons: Ah, nifty.
Yeah, I'd like to work with a Juniper sometime.
We had one in the lab at Cisco but nobody knew how to use it and it wasn't hooked up.
up_the_irons: haha, Cisco had some Juniper gear in a closet, nice
mhoran: Heh. Yup.
Fuckign monit. What a shitty piece of software.
up_the_irons: oh certainly, monit is garbage
runit is the way to go
***: cablehead has quit IRC ("Leaving.")
mhoran: Interesting.
up_the_irons: runit has one job -- to keep a process running
mhoran: Heh.
up_the_irons: you can trust it to do that
***: cablehead has joined #arpnetworks
mhoran: Does it just do pid checks or can it do more advanced things like HTTP?
Our AOLservers like to die.
up_the_irons: mhoran: no, pid check is the way *not* to manage a process, because that's akin to "ps auxw | grep <PID>", subject to race conditions, etc... It manages a process because it is the parent of that process. Only in that way can you fully control a process
mhoran: Nifty.
up_the_irons: then controlling the process is easy, like:
"sv stop my-proggy"
sv start
sv status
etc ...
mhoran: Ah, right.
If I dump monit, I still need a way to be alerted of service restarts, and to do those HTTP checks and reboots.
up_the_irons: and you can set groups that have access to individual supervisors, so there can be a group "web-admins" that can control *this* and "dns-admins" that can control *that*
mhoran: That's nice.
up_the_irons: mhoran: yeah, runit won't focus on those as much
mhoran: Still, looks pretty solid.
up_the_irons: i run it (pun!!) on a lot of boxes, never was disappointed
mhoran: Of course, if runit dies, you're fucked.
:(
(I guess if init dies, I'm screwed too.)
toddf: cronjob a check to confirm runit is running
mhoran: Yep.
monit could easily be replaced with some cronjobs and runit.
I like it!
up_the_irons: mhoran: you set runit to start in /etc/inittab (old way) or /etc/event.d (newer Debian / Ubuntu)
mhoran: so it always stays running
mhoran: Right.
up_the_irons: even if you kill it, it'll restart
mhoran: Yeah.
up_the_irons: it's very "unixy", look at the cool stuff I can do:
$ for i in host1 host2 host3; do ssh $i sv status vm-management; done
run: vm-management: (pid 11031) 133590s; run: log: (pid 12944) 1926719s
run: vm-management: (pid 10962) 133570s; run: log: (pid 21526) 205819s
run: vm-management: (pid 4438) 124502s; run: log: (pid 4437) 124502s
$
(host names changed to protect the innocent ;)
mhoran: That's nice.
up_the_irons: cd $office
toddf: no wonder my vm came up quickly when you killed it the other month (at my request of course)
up_the_irons: toddf: did you try the new console? you can now kill and start your VM yourself ;) Not that I expect you to jump at that option...
toddf: I will, but not yet, dealing with a cap growth issue at my parents house in their economy puters
up_the_irons: roger
jeev: http://cgi.ebay.com/1U-Super-Server-X8STi-F-Core-i7-920-DDR3-1333-8GB-RAM_W0QQitemZ120407077979QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item1c08d2345b
what a trick
that's just EVIL
up_the_irons: what trick?
***: cablehead has quit IRC ("Leaving.")
cablehead has joined #arpnetworks
cablehead has quit IRC (Client Quit)
jeev: showing all that but only selling the barebone
***: cablehead has joined #arpnetworks
up_the_irons: jeev: hah, yeah
jeev: it's mobo and case i guess
up_the_irons: yeah
jeev: thing is, i don't see all too much the value add for screwing in the MB ;) you can buy the case and MB separate, and save money
***: vtoms has quit IRC ("Leaving.")
jeev: yea
eww
my kvm box oopsed on me!
general protection fault: 0000 [#2] SMP
gizay
it was working with 2.6.32-rc3
up_the_irons, you know what KSM is
up_the_irons: jeev: nope
jeev: apparently it makes it possible for you to be able to run more guests than before
http://fedoraproject.org/wiki/Features/KSM
up_the_irons: jeev: i imagine that would be an overall goal anyway
jeev: how many do you think i could stuff on a E5420 with 8 gigs of ram
just curious, probably gonna use it to set up a mail server for something unrelated to my shit
up_the_irons: CoW memory across VMs, yeah this concept has been done before
jeev: obviously, that depends on how much RAM you give to each VM
jeev: yea i know but CPU wise
up_the_irons: depends how much load you put on each VM
jeev: :>
up_the_irons: there is no general answer
***: nuke^ is now known as nuke
-: sroute runs for the bomb shelter
***: dbgi has quit IRC ()