***: schmir has joined #arpnetworks
schmir has quit IRC (Remote host closed the connection)
schmir has joined #arpnetworks dxtr: Yay! My throat feels like its been stabbed and I'm coughing my guts out bob^^: swine flu dxtr: Impossible! My school gave me the flu shot!
THIS IS MADNESS!
I'll sue them bob^^: they *told* you it was a flu shot
really it's a government research project dxtr: Probably
I don't care - it went quick and (not really) painlessly
Hey, RandalSchwartz. You there? cedwards: that's why I don't bother with flu shots. Still no guarantee you won't get the flu. ***: heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer
ziyourenxiang has joined #arpnetworks
lll has quit IRC (Quit: ZNC - http://znc.sourceforge.net)
lll has joined #arpnetworks
amdprophet has quit IRC (Read error: Connection reset by peer)
amdprophet has joined #arpnetworks
fink has joined #arpnetworks
LT has quit IRC (Quit: Leaving)
ziyourenxiang has quit IRC (Quit: ziyourenxiang)
st3ffan has quit IRC (Quit: leaving) nuke-: anyone around from staff? Wraithan: up_the_irons: ^ nuke-: idle 17h :|
:p RandalSchwartz: depends on which staff.. :) nuke-: arp staff
:p dxtr: Hey, RandalSchwartz! :)
http://github.com/dxtr/emacs/blob/master/init.el <- My current emacs setup! ***: nerdd has joined #arpnetworks
nerdd_ has quit IRC (Ping timeout: 240 seconds)
sbp_ has joined #arpnetworks
sbp_ has quit IRC (Quit: Lost terminal) fink: emacs = sweet Wraithan: my emacs is pretty basic fink: vi* sux mike-burns: Hey watch it. Wraithan: http://sprunge.us/FNBY -: Wraithan uses vi(m) often jdoe: haha -: jdoe starts more holy wars jdoe: vi* > emacs! perl > python! linux > bsd! Wraithan: Most times I have emacs but quick edits, especially on work servers (where I don't have an emacs daemon running), I tend to use vi(m) fink: jdoe: wrong on all counts jdoe: fink: yhbt. yhl. hand. -: jdoe lights a cigarette and basks in the afterglow fink: Wraithan: ed is better for quick edits, imo mike-burns: I write Ruby in vim all day. It gets the job done. fink: jdoe: lol Wraithan: fink: if I knew ed maybe. fink: Wraithan: ed's a nice guy Wraithan: friend uses joe mike-burns: Yeah, I like ed for quick edits. Wraithan: fink: ed sits next to me at work RandalSchwartz: but joe uses ed. :) -: jdoe ONLY USES PICO Wraithan: fink: though he prefers Edwin jdoe: ... not these "open source" knockoffs like nano >:| RandalSchwartz: Emacs can emulate *all* of your editors. :) jdoe: and your browser.
and your irc client. fink: nano > vi* jdoe: and your os. Wraithan: notepad.exe > * -: fink has a soft spot for pico RandalSchwartz: cp /dev/null notepad.exe Wraithan: RandalSchwartz: hater.
:( -: Wraithan goes back to work Wraithan: in emacs jdoe: ... nah I don't actually care. I just learned vim first and have no desire to remember another trillion or so key bindings. -: RandalSchwartz goes back to reading irc... in emacs
fink goes back to emacs… in emacs jdoe: RandalSchwartz: ... yeah? Impressive.
I know you can do it, I didn't know anyone actually did. RandalSchwartz: that's how I'm typing right here right now jdoe: neat. RandalSchwartz: Terminal.app -> ssh -> screen -> emacs -> irc.el -> freenode Wraithan: I have some friends that use emacs for IRC jdoe: die hard irssi user. Wraithan: weechat here
weehchat > irssi jdoe: konsole -> ssh -> screen -> ssh -> screen -> irssi :P Wraithan: because python + perl > perl jdoe: Wraithan: the nicklist is neat, but it was unstable (like crashy unstable) last I tried. Wraithan: urxvt -> ssh -> tmux -> weechat -> freenode
btw
tmux > screen
and urxvt > * jdoe: yeah, uh, what's so great about tmux? Wraithan: jdoe: weechat is a lot more stable now a days, though I run against head jdoe: I saw they rewrote it, I didn't see why
er, rewrote screen Wraithan: jdoe: eh, there isn't much that makes tmux better, I like the defaults better and it is on all of my systems
It is part of the set of packages I have my default install jdoe: ah, fair enough. RandalSchwartz: It's BSD licensed, not GPL jdoe: this is the only obsd system i've touched in years, so I think I'll be sticking with screen RandalSchwartz: that's part of the reason fink: Wraithan: does tmux let you have tmuxes inside tmuxes? jdoe: RandalSchwartz: I guess, but most of the software people are going to put on the machine is going to be some non-bsd license. Wraithan: fink: you can force it yes RandalSchwartz: and client/server design, which lets it get away with a few things Wraithan: fink: it gives a warning just like screen does fink: ok RandalSchwartz: jdoe really? very little of what I have on my box is GPL Wraithan: BSD licence is part of what makes me consider BSD
Would be nice to get away from the GPL a bit more RandalSchwartz: replacing gcc is the tough one Wraithan: clang is getting closer every day RandalSchwartz: but there are new compilers, I hear Wraithan: can't you build the whole freebsd userland with clang now? including the kernel
plus clang is LLVM based, something which I support (even if it is funded by Steve Jobs.
Thought they are getting closer to compiling the linux kernel with it as well, which is cool jdoe: RandalSchwartz: apache?
RandalSchwartz: perl? ;) RandalSchwartz: what about apache and perl?
neither of them are GPL jdoe: you said GPL, I said non-BSD.
perl is dual-licensed, right? I skimmed over the terms, the "artistic license" doesn't seem far off the GPL. -: jdoe shrugs. jdoe: emacs is gpl, if you use a real mail client it's probably gpl'd (or worse) as well... mutt is GPL, alpine is GPL... CESSMASTER: gpl OR WORSE
dun dun dun Wraithan: yeah... the emacs thing, not sure what I would replace it with jdoe: CESSMASTER: well, pine is fairly restrictive, that was what I was thinking of. cedwards: anyone good with network bridging and routes? Wraithan: cedwards: I've done it... in some moderately complex setups but I wouldn't claim to be good (aka I spent hours with man pages and google to do the harder ones) ***: nbari|away has quit IRC (Ping timeout: 240 seconds)
nbari|away has joined #arpnetworks dxtr: Hmm... I'm having problems with emacs now cedwards: I'm getting my hours in now..
I have a kvm host. I need 9 vm (which are created) to connect through two vlans. I assume this needs two bridges.
what I'm currently trying is eth0 (host), br0 (half guests), br1 (half guests). up_the_irons: cedwards: when you say vlans, do you mean real 802.1Q vlans, or kvm's mis-use of the word vlan ***: Wraithan has quit IRC (Quit: rebooting)
Ehtyar has quit IRC (Quit: Don't follow me)
Wraithan has joined #arpnetworks
fink has quit IRC (Quit: fink)
schmir has quit IRC (Remote host closed the connection) jdoe: up_the_irons: might want to specify the zones for rdns in the faq. You're using a hyphen for the mask, RFC uses a /
er, zone style up_the_irons: jdoe: thought the RFC showed both styles
but yes, i should specify jdoe: rfc only has examples for / but I think it says you can use whatever delimiter you like. up_the_irons: jdoe: ah ok jdoe: er... wait
is this even right?
242.97.136.174.in-addr.arpa. 3600 IN CNAME 242.240-247.97.136.174.in-addr.arpa.
shouldn't that be IN CNAME 242.240-29.97.136.174.in-addr.arpa
up_the_irons: well, whenever you're around, ^^^ :) up_the_irons: jdoe: no
jdoe: use the format I specified
it's not simply s/-/// jdoe: mmk.
fwiw, that's not how 2317 does it either. up_the_irons: so you don't think I'm nuts, that format was specified in "DNS & BIND", 2nd Ed. by O'Reilly (old news now) jdoe: right on.
I mean, it doesn't matter either way, you could call it bob.in-addr.arpa if you wanted. up_the_irons: right jdoe: alright, I'll fix the zone too. Wraithan: up_the_irons: that book is sitting on our shared bookshelf at work up_the_irons: Wraithan: hehe, nice. the 2nd ed. is old now, but that was current when i read it jdoe: I have it at home somewhere. up_the_irons: i think everyone has a copy somewhere jdoe: not sure why given that I don't generally use bind. Wraithan: I am sure it was new when my boss bought it... we have lots of old books jdoe: I did learn something new from it though, DNAME Wraithan: I've never even opened it up_the_irons: whoa, wth is DNAME jdoe: haha.
sorta like cname but for an entire subdomain up_the_irons: whoa jdoe: I believe it's new in BIND9
"new" ***: amdprophet has quit IRC (Ping timeout: 240 seconds)
amdprophet has joined #arpnetworks dxtr: Is someone here using wanderlust? cedwards: up_the_irons: I mean 802.1q vlans. our office network is separated by business unit per vlan.
up_the_irons: I need one kvm guest to go to one vlan, and others to another. up_the_irons: cedwards: so the first thing you'll need to do is set up VLANs on the regular interfaces (eth0, eth1, ...)
cedwards: or, unless you have one vlan going to eth0, and another to eth1, etc... ***: visinin has joined #arpnetworks jdoe: visinin: nice work. visinin: jdoe: what jdoe: visinin: ident
er... non-ident. visinin: oh yeah, thanks :3 cedwards: up_the_irons: currently the setup is eth0 connects to a port on the switch, vlan24. eth1 connects to a port on the switch vlan64.
up_the_irons: I had set eth0 to br0 and eth1 to br1, but I'm thinking I should dedicate an interface to the host, leaving me with eth0, br0 and br1.
up_the_irons: eth0 connecting physically to one vlan on the switch, br0 and br1 connecting to their respective vlans on the switch and the guests assigned to each bridge accordingly. up_the_irons: cedwards: so you would make eth1 a 802.1Q port? ("trunk port" in cisco terms) ***: schmir has joined #arpnetworks cedwards: up_the_irons: is that what we need to do? we did consider trunking, but hadn't tried it yet.
up_the_irons: the server I am using has 4 NICs, and between the 9 VMs + host we only need to access two 802.1q vlans. up_the_irons: cedwards: you don't *have* to but then you'd have to pick vlan 24 or vlan 64 for the host
cedwards: oh, you have 4 nics cedwards: up_the_irons: right. with the 4 nics I was directly connecting to ports configured on the appropriate vlans. up_the_irons: cedwards: well, easy solution is to bridge vlan24 VMs to eth0 and bridge vlan64 VMs to eth1, for example cedwards: up_the_irons: the problem arose when I added the second bridge on the second vlan. the routes got confused I think and I couldn't get anywhere. up_the_irons: and use eth2 for the host, if u wish
your setup would look something like:
br24 bridges eth0, tap1, tap2, etc...
br64 bridges eth1, tap6, tap7, etc...
i'm making up the interface names cedwards: would this be an appropriate config: http://pastebin.com/ZRqNAvmA
I'm just making up addresses there.. up_the_irons: cedwards: looks about right cedwards: up_the_irons: would I need to do anything additional with my routes/gateways?
(this is all on ubuntu 10.04 server, btw) up_the_irons: cedwards: not really, you're just bridging your VM traffic onto the wire and not using the hosts routing, I'm assuming
(which is the better way to go IMO)
*host's cedwards: I'll have to keep playing with it tomorrow.
I think my initial problem was the host interface was also the br0 (eth0->br0), and br0 address was how I accessed the host. up_the_irons: cedwards: in your eth0 stanza of your paste, remove the "bridge_" directives
that only applies for a "brXX" interface CESSMASTER: the eth0 stanza
hahahaha cedwards: ohh, right. copy/paste blunder there. up_the_irons: yeah i figured cedwards: I'd try this out here on my laptop, but no bridging over wifi and I only have one nic. -: Wraithan isn't sure what is funny about eth0 stanza up_the_irons: cedwards: when you bridge eth0 to something, you want the IP addresses that originally existed on eth0 to now be on the bridge. so you'd remove the IPs from eth0 -: up_the_irons isn't sure what is funny either CESSMASTER: is "stanza" commonly used to describe conf files?
I only see it for poetry
in which context it is hilarious Wraithan: CESSMASTER: not super common but it is used for that cedwards: up_the_irons: yeah. when I bridged eth0 to br0 I gave br0 the IP. In my example, eth1 and eth2 will not have IPs, only br1 and br2. up_the_irons: CESSMASTER: the "iface" parts are called stanza's, i believe
cedwards: ok CESSMASTER: well I bet it was hilarious when somebody first came up with the name :/ cedwards: ..which does make me wonder, if a bridge is basically just a software switch (in simple terms), why would it need an IP at all? up_the_irons: cedwards: management cedwards: does it _need_ an IP? I mean, if I can manage them via eth0.. Wraithan: CESSMASTER: beer and IT seem to go hand in hand... up_the_irons: cedwards: if you don't have any IPs on your bridges, you won't be able to access the host machine (only VMs). Unless you dedicate a separate port for management (which I think you are doing)
cedwards: no, bridges don't _need_ IPs
i tend to not put IPs on mine cedwards: In my mind a clean setup would be non-bridged eth0, with a static address. then non-addressed bridges with VMs attached, each having their own static addresses.
bridge management can be done via eth0 (host), and it serves the main purpose of the VMs getting in/out traffic.
I can then create as many vlan-attached bridges as I need, without allocating any IPs to the bridge interface itself.
(vlan-attached meaning physical interface connected to vlan assigned port on switch) up_the_irons: sounds fine to me ***: fink has joined #arpnetworks
fink has quit IRC (Read error: Connection reset by peer)
schmir has quit IRC (Remote host closed the connection)
visinin has quit IRC (Quit: leaving)
amdprophet has quit IRC (Quit: amdprophet)
amdprophet has joined #arpnetworks jdoe: ... that's kind of... weird.
chrooting postfix usually involves some work.
... on openbsd it was just telling it to in master.cf, no adding files to the chroot.
doesn't seem right. up_the_irons: nice jdoe: haha. I guess. I suspect it's broken until proven otherwise. The fact that mail's coming through isn't swaying me in that regard. up_the_irons: ah ***: heavysixer has quit IRC (Quit: heavysixer)
heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer
fink has joined #arpnetworks
heavysixer has quit IRC (Quit: heavysixer)
ballen has joined #arpnetworks
ballen has quit IRC (Changing host)
ballen has joined #arpnetworks
ChanServ sets mode: +o ballen
ballen has quit IRC (Quit: Leaving...)
heavysixer has joined #arpnetworks
ChanServ sets mode: +o heavysixer infrared: qmail++ -: infrared hides Wraithan: s/q/g/ infrared: hah mike-burns: I'd much rather set up qmail on my server than gmail. Imagine what's involved in running gmail! infrared: mike-burns: a few thousand servers and a bunch of people with doctorites running around screaming python mike-burns: Exactly; I don't think I can afford that. infrared: i think google jumped the shark
idiots mike-burns: They did that years ago. infrared: i dunno.. i really think this was the topper for me mike-burns: What are you going to switch to? infrared: nothing. they own me
bing? and stare at that horrible home page? mike-burns: Well you'd use a smart bookmark/search bar to never look at the home page, but yeah---Y!, Bing, and G all have identical search results HTML now.
And Ask censors their results.
There's Teoma. Wraithan: Gmail is Java isn't it? infrared: probably Go or something stupid mike-burns: I'd expect it to be C. infrared: who in their right might would code a web app in C?
mind* mike-burns: OKCupid, Google. I think that's it. Wraithan: A lot of the backend stuff at google is done in C, and some more recently is Go mike-burns: Lots of the github backend is in C. Wraithan: But they admitted in one part of their FAQ that gmail is java iirc ***: ballen has joined #arpnetworks
ChanServ sets mode: +o ballen infrared: java-- Wraithan: Agreed mike-burns: http://panela.blog-city.com/python_at_google_greg_stein__sdforum.htm - "Most other front ends are in C++ (google.com) and Java (gmail)." ballen: oooh hi CESSMASTER: amazon's shit used to be coded in C infrared: i thought amazon was perl Wraithan: CESSMASTER: amazon's shit used to be written in Lisp mike-burns: Yeah. CESSMASTER: obidios or whatever was called was written in C, iirc Wraithan: Reddit was in lisp
now it is python
iirc mike-burns: So many things can compile down to something that gcc can understand that it doesn't make sense to write C anymore. Most of the time. Usually. infrared: haskell Wraithan: Like the haskell kernel module folks
lol infrared: haha
jinx Wraithan: infrared++ mike-burns: Yeah, I had Haskell in mind. infrared: i dunno.. i still have a love for perl
:) mike-burns: They're doing some LLVM work on GHC or something, supposed to be faster and more memory efficient. Wraithan: infrared: we are all wrong sometimes. -: infrared punches Wraithan in the juggular Wraithan: hah ***: dreamer1101 has joined #arpnetworks dreamer1101: hello, is there anyone from arpnetworks available to spare a couple minutes? Wraithan: up_the_irons: ^ ballen: not from ARPNetworks, but whats up dreamer1101: oh, well i placed an order a couple days ago and haven't received the verification call... i'm currently out of town and was wondering what is up... was kind of wanting to setup the base of my system before i head to the beach this weekend... RandalSchwartz: did you place the order after the orders were closed?
or are they back open again? ballen: still closed atm Wraithan: dreamer1101: I only got a verification email dreamer1101: oh ok cool
well i've been in contact via email with gary from arpnetworks and he provided me with a link to order... and i did =) ballen: there ya go dreamer1101: just haven't received a phone call or email back after i placed the order
hehe
:D Wraithan: dreamer1101: you talking the special order link? ballen: seems quite unlikely you'll get a call dreamer1101: just an email providing further details of what i ordered then i assume? ballen: yep
I assume Gary is just busy Wraithan: dreamer1101: along with a recept once your card is charged dreamer1101: yea i totally understand and not here to rant, i just found it interesting when i clicked the contact us link that arpnetworks had an irc channel, i was going to send a msg- but like you said, he is prolly just a bit busy lately... ballen: ya dreamer1101: cheers everyone
i've been in a vehicle for 14hr today (i know airplanes exist) but... ya know ;) ballen: hah that sucks dreamer1101: thx for the response guys ballen: np dreamer1101: bye
have a good night fellas ***: dreamer1101 has quit IRC (Quit: leaving) cedwards: nite fink: paste
sorry jdoe: whoop, dns is all happy and good, thanks up_the_irons. fink: jdoe: what was your issue? jdoe: fink: well, first I hadn't set it up. Then I had, but due to some miscommunication it was setup wrong. Then it was setup right ;) fink: jdoe: just wondering if it was something i should take notice of with my vps jdoe: fink: not unless you want him to delegate rdns and you haven't set it up, or you have set it up and it isn't working. fink: no i think it's working ;) thanks to tinydns ballen: rdns delegation == pita
especially classless jdoe: ballen: yeah. I'm not thrilled with the RFC for it.
... because it only suggests how one might do it.
it doesn't set a standard. ballen: hah jdoe: as evidenced by our back and forth a couple hours ago. ballen: ya ***: jdoe_ has joined #arpnetworks
jdoe has quit IRC (Quit: whee.)
jdoe_ is now known as jdoe jdoe: whee.
buh bye dedicated. ***: fink has quit IRC (Quit: fink)
steinberg has quit IRC (Quit: leaving)
ballen has quit IRC (Quit: ballen) up_the_irons: jdoe: cool, glad to hear
I think I will start putting /64's (IPv6) on the wire for new orders. Everything the same except /64 instead of /48. The /48 will still be allocated to the customer, but it won't do anything unless one contacts support and wants it routed over a link-local
toddf: ^^ if you have an opinion, lmk
then I won't be a "bad example" jdoe: why change? up_the_irons: jdoe: a /48 for a site is meant to be subnetted further for different purposes. But the way I currently set this up, it is not possible to subnet further. No one besides toddf has ever even asked to subnet it further, so I think it is clear that more than one subnet is not required by default. one subnet = /64. and if more than one subnet is required, all that is needed is to contact
support and we'll route it over link-local. it's like sensible defaults. right now, my default is wrong jdoe: faq entry in the making, huh? ;)
yeah, that makes sense.
even a /48 is insane. up_the_irons: hehe, yeah. but a /48 is "generally accepted" as default allocation for a "site"
those clinging onto IPv4 thinking will still say don't allocate a /48 if it is not needed, but I prefer the new thinking of "everyone gets the same thing", regardless of need. If you do the math, we won't run out of subnets, so it doesn't matter. the upside is it makes allocation policies a lot easier, and therefore cheaper to maintain jdoe: true. ***: heavysixer has quit IRC (Ping timeout: 260 seconds)