brycec: Webhostbudd : glad you guys worked it out :) up_the_irons: why on earth does that ip exist.... im just curious why i got the routing loop no route to destination, router sends it out default route, at which point the default route sends it back, etc... another option is to blackhole it, but not all routers are config'd that way hmmm shoulnd't it send the icmp for no route and then traceroute stops Webhostbudd_: no, because it does have a route, the default only if i set a blackhole route giving it essential no destination would that icmp for no route be sent; at least i think that's how it works heh inspiring confidence as usual, up_the_irons :P I wonder if there is a non-graphical way to check my current month used bandwith? e.g. by text or bare html? vnstat vnstat ? vnstat http://freecode.com/projects/vnstat but... vnstat measures on your side, not up_the_irons' so their could theoretically be a discrepancy. *there correct. especially if you chatter between two or more vm's Great point toddf - vnstat will count ALL network chatter, even if it doesn't hit the Internet. up_the_irons only measures what hits the Internet. you might even correct that further up_the_irons currently only meters what goes across his IPv4 router wha??? really??!? IPv6 traffic is currently unmetered because it actually goes across a totally different router Nice loophole, up_the_irons :D and he knows this and has acknowledged it and decided to not do anything about it as a benifit of IPv6 usage trust me I utilize this in ways I'd prefer not to admit <3 IPv6 I will too... when I migrate my VPS data over to ARP :D Now if only Amazon S3 had IPv6 endpoints... something tells me you won't be using 2 gif(4) interfaces on openbsd to permit af-to inet6 in pf.conf to send v4 afs client requests over v6 to your afs servers elsewhere serving files for your web pages .. ;-) hahaha genius, toddf you can just af-to inet6 for v4 traffic you originate (note not that you respond to or it'll not work) and go out another location that doesn't have bandwidth caps (like home business class cablemodem for example) how do I know how much bandwith lasts for this month? maybe that can be done? warex: if you don't have a bandwidth thing listed on your services you'll need to send a support@ ticket to request access to your graphs, last I understood at one point not all vps customers were setup with graphs since not all customers even looked at them now if only I could figure out how to get asterisk to listen on v6 for iax and sip I'd be golden ;-( $ sudo grep bindaddr /etc/asterisk/sip.conf bindaddr=[::]:5060 bindaddr=0.0.0.0:5060 $ fstat | grep 5060 _asteris asterisk 10915 14* internet dgram udp *:5060 meh ;-( toddf: isn't there a sysctl setting that sets the default bind to be ivp6+ipv4? At least, I know there's one on Linux... Pretty sure Asterisk is ipv6-friendly too, but can't swear to that. (-friendly, meaning it should happily bind) IPv6 <3 asterisk is... sorta ipv6 friendly. not all protocols will handle it, iirc. ... though I'd expect sip/iax2 to. I would too... Then again, there's a distinct lack of ipv6-friendly phones so I've never tested it out. Really a shame too - if there's one single application that will immediately benefit from the NAT-less world of IPv6, it's SIP traffic. you will find the sysctl that 'binds to ipv6+ipv4' to be disabled and code missing on openbsd. it is a security issue afterall. any app that has a clue binds to one socket per af. I don't need an ipv6 friendly phone. I can af-to inet6 from iaxcomm (softphone on openbsd) to the asterisk box. unfortunately I've not found the reverse to work as I'd expect Agree it does present a security issue and hence automatically excluded from OpenBSD There are softphones, but I don't use or deploy those in a professional capacity (for many reasons, the least of which being they always seem to behave poorly, crashing etc) toddf: lol, that *is* genius ;) brycec: what, the ipv6 loophole? up_the_irons: yeah, ipv6 is un-billed (according to toddf) (And we can all trust what toddf says, he's got +o :P) when total IPv6 traffic across my entire network isn't measured in "Kbps" (yes, "K"), then maybe i'll care about metering... ;) i shouldn't be so harsh, at this point, it's probably like 2mbps brycec: yeah ipv6 is not metered and now i see 80 mbps of ipv6 traffic Very cool. 4 billion IP addresses and all the traffic I can come up with. This smells like a good time for some routing loops guys, my IPv6 router is a small OpenBSD box with 100 Mbps total, please be nice to it Yep, that's probably me... :D oh gee, only 100Mbps on it? poor thing I'll be done soon. Doing some semi-legit throughput testing. brycec: if you could throttle that i'd appreciate it; i don't want IPv6 to get slow for others brycec: roger it's iperf... throttling defeats the purpose :P but I'll shut it down very shortly ah i c The question and quiz today is: What does Bryce's other (soon-to-be former) provider do about prot 41 traffic? so I'm just cranking some out... then I'll look at their usage graph Funny enough, all that stemmed from "why is my backup taking so long to transfer?" up_the_irons: Do you have any form of traffic shaping, fair-use, QOS etc in place for IPv4 traffic? (I'm assuming not for IPv6 based on your earlier statements) Okay, I'm done. The b/w is all yours, #arpnetworks brycec: tnx! brycec: we don't really have that stuff for IPv4, i want to provide as clean a link as possible. there's the SSH SYN rate-limit (inbound) and then 5 Mbps UDP max on outbound (which i don't care to remove if someone needs it, but nobody seems to) what do you guys think should i ditch my shared hosting provider for small personal projects and get a cheap arpnetworks vps? Well obviously you should send #arpnetworks money :p Really though... Boils down to the value of a server (colo, self-hosted, or a VPS) to you vs a shared host's "hands off" approach hah Personally, i have a cheap $6/mo shared host with "unlimited" bandwidth and "unlimited" disk space where I host most of my websites. They aren't mission-critical, the reliability seems 100%, I don't have to maintain the host, patches etc, and I know that me dicking with firewall rules isn't going to knock a client offline. everything about that seems to be awesome for you But I have two VPS for more "developery" things... Run it all myself, tweak things, manage and oversee my own VPN network, etc. I can handle downtime if it's my fault :P That's the sort of thing a VPS is perfect for, IMO heh yes please guys, stop 'testing the waters', I like my nice and crisp v6 behaviors ;-) haha jdeuce: i think you should get a cheap ARP Networks VPS, but i'm biased ;) brycec: heh I was obviously reading dated scrollback... obviously of course, I read and reply to dated scrollback. As does up_the_irons. damnit ppl stop ordering vps', i can't get anything done.. ohwait... ;) brycec: are you tunneling? :) mercutio: eh? in what sense... you were saying about protocols heh it's jdoe (or toddf? don't remember) who's tunneling I mean, i tunnel... but not my my ARP VPS ahh you were saying about qos only use for GRE tunneling I have is for IPv6 connectivity (e.g. my HE tunnels), but up_the_irons is super awesome and provides native ipv6 err what your provider did to other protocol traffic ahh now I'm getting confused by mercutio's confusion heh ok i think it's me that got confued protocol 41 is ipv6 yes protocol 47 is gre you're reading the backlog too closely now :p no i just noticed it before i said anything Mostly I was wondering what, if anything, prevented a user such as myself from hogging all of up_the_irons' bandwidth (as I had done previously with IPv6 traffic) well mercutio I said 41 when I meant 47, so there :P brycec: haha ok i use ipip for tunneling it's basically 20 byte (ip size) overhead proto 41 is known in bsd land as gif(4) incase that helps oh isn't proto 41 plain IPv6, no encapsulation? twobithacker: yeah you have the IPv4 overhead then you have raw v6 but if you add a gif(4) interface to a bridge(4) interface it adds ethernet overhead as well ah, gif can use etherip oh joy. asterisk can do IPv6 for sip. it just can't do it at the same time as IPv4 on OpenBSD. sucksucksuck. hahaha someone probably assumed you'd use the net.ipv6.bindv6only :P Oh assumptions... sorry, BSD would be net.inet6.ip6.v6only let me see if hitting peers via the v6 mapped v4 address that another router is af-to inet'ing back to v4 that sooo broke my head then you do not wanna see my pf.conf that does af-to from v6 to v4 and in other cases v4 to v6 and in other cases 'pass in inet/inet6.. rdr-to ip tag foo' 'pass out inet/inet6 nat-to ip tagged foo'