kvr03 going down in 8 minutes kvr03 is back up and VMs are coming online heh hi =D not that I'm complaining but is the bandwidth meter working since a couple weeks ago? all my datapoints are empty for the daily and weekly charts, right above week 20 it goes empty mine look ok i think http://i.imgur.com/oLr5HpR.png oh, hm week 20? http://i.imgur.com/lu4QE7k.png this looks odd toddf: hazardous : anyone switched to the 1 gbps port won't have anything in their graphs (known issue) hazardous: indeed week 20 looks weird ;) oh, didn't realize that was an issue <3 Now I feel self-conscious about how little b/w I'm using :( <1GB in the last month I know! I'll just iperf around the net a bit, chew up my quota up_the_irons: have many people switched to the gigabit yet? mercutio: a few brycec: lol did anyone notice a diff? what - there's something new to try here? :) :o RandalSchwartz: i think you have to make a request before up_the_irons goes to datacentre although maybe he's plugged in all the hosts to it now mercutio: nope, it's all done :) but from my understanding, some people are having their /30 shifted to the new router, which terminates on gigabit rather than 100 megabit up_the_irons: ahh cool mercutio: i can re-terminate vlans to the new switch all remotely hmm but i have bgp so it's complicated for me :( make a request about what? where would I have learned abut this? oh! mercutio: yeah i know, still need to figure out what to do about that RandalSchwartz: only by lerking in here. i haven't officially announced anything. ok would I have to have new settings for my /etc/rc.conf? like for ipv6? mercutio: first thought would be to open 2nd session on new router. those with bgp should actually desire 2 sessions if they want true redundancy. i'd announce a default route downstream from both. in this case, bgp customers would have vlan active on both switches.. i guess... gotta figure that out still (don't want to create a loop) RandalSchwartz: no changes necessary on your side but of course all this reminds me that before I can install my next package, I have to upgraded from perl 5.10 to something that is actually supported. :) up_the_irons: i assume you don't want to allocate another vlan to the other switch shoemaker's children and all that i assume with one vlan the way to go would to have the vlan not joining between the two routers. mercutio: that is a possibility actually. with new IP block. hmm there's also the possibility of using /31s. mercutio: indeed, can't join between routers, but on vm hosts there is a bridge. need to test if this would be a problem. i dunno if you'll run out of ip's or vlans first. probably IPs :) damn this ip starvation esp. if I get a 2nd location going and put a /22 or something over there up_the_irons: if you're only putting a /22 in second location, then testing /31s is probably prudent. well if wanting to ahve more than 250 customers in other location that said, if you have two locations then having a /24 anycast between the two would be cool. no idea how much demand though and it's a bit complicated if people want proper redundancy for it mercutio: i have had a couple rare inquiries for anycast services. good thing about anycast is you can charge a lot ;) up_the_irons: heh yeah ppl pay big bucks for that kind of redundancy interesting. i suppose i'm more of an engineer than sales droid. i see it more as a cool thing to do :) it looked liek the second location may be a joke before when i read about it something about some holiday checking out datacentres. but it's pretty likely ? i also think of it more of as "a cool thing to do"; getting paid for it is just a nice side effect ;) the 2nd location is no joke, but it's also not close. very likely a 2014 endeavor. i found some kind of limited bgp daemon without forwarding that made it easy to do any cast type things before can't find it now though ahh ok well it's already half way through 2013, and xmas seemed liek a few weeks ago :) yup srsly i think i found something diff from last time heh ahh found the original thing exabgp basically it makes it easy to do things like withdraw route when a daemon isn't, inject it when it is etc which means if you have a tunnel between multiple locations, or multiple hosts within oen location should be able to have multiple hosts serve the same ip for a service and just advertise it as a /32 of course there is the bgp complication heh, i don't think i've ever seen so many security vulnerabilities in a row.. the X library stuff effects heaps of packages on ubuntu yes my inbox was flooded with DSAs same here oh it seems exabgp has an ubuntu pacakge oh, freenode is being ddos'ed again