guess people liked that one better i'm not actually following that one More twerking. i still don't really get twitter well i mean i know how it works, but it seems like just another thing to try and keep up with the nlnog ring looks like a really nifty concept mercutio: I have bitlbee logging tweets and sometimes I'll search for a given topic, but there's no way I'd read things as they come in, too much info, to much attention and time wasted .. ;-) avj: the ring is awesome very nice for diagnostics unlike route servers, you have real tools, like mtr i like things like mtr :) i've noticed that cisco etc route servers are always slwo and don't haev nice tools :) https://ring.nlnog.net/ Oh right because ARP's a participant. i was also wondering what the bgp collective was i went to the page i didn't quite understand if they were going to bridge different areas or what ok this ring thing does look cool. So essentially this is a machine (VM) on your network that you allow other networks' admins access to (and in return, you get access to theirs) in order to troubleshoot the Internet? Is that essentially right? that's what the web page suggests i wonder there are special mesh traceroute tools there are ring scripts to do multihost https://ring.nlnog.net/user-guide/ looks good i think it ssh's to every host in turn which could be slow :) in turn? ick I would hope it would be parallel maybe it is, but it doesn't look like so maybe up_the_irons can confirm when he's aorund :) https://raw.github.com/NLNOG/nlnog-ring/master/scripts/ring-ping yeh it ssh's to each one and there are over 200 oh i think the idea is you go to a susbset of htem Actually it looks like xargs does the parallelization (but that's just a quick skim oh i think you're right i hadn't worked my way down that far -P10 -- run up to 10 procs at a time i was still trying to figure out how it gets the hostnames out of dig Fun with reformatting ;) (ignore comments, strip double quotes, and convert spaces into newlines) SERVERS=$(dig -t txt +short ring.nlnog.net | grep -v '^;' | tr -d '"' | tr ' ' '\n') Man that's some clever parallellization voodoo heh so 237 hosts bash isn't known for being good at parallellization... but here it is molded to do it :) 10 at a time with 2 second connect timeout so it probably still takes 20 seconds to run or something but not so bad Better than 4 minutes parallelisation is annoying :) yeah i'd rather see main hubs that send the commands off to servers adn give the responses back and maintain persistent connectinos open to them but i imagine it'd take more effort to implement than the amount of time you'd save err take more time to implement More moving parts to break too And eats up bandwidth (not much, but it adds up) in some ways, it also means can throttle requests or such easier True if someone has a script error or something that ddos's servers accidentally or something i doubt bw would increase noticably because you also remote ssh connection initialisation overhead etc so oyu may send 250 bytes instead of 2k for persistent versus creating new connectoin plus less cpu Depends how much the ring is used probably smokeping takes up more than manual testing You can do them one at a time if you want https://github.com/NLNOG/nlnog-ring/blob/master/scripts/ring-all-serial if you're smokeping'ing lots of hosts you can reduce bw slightly be reducing the ping size heh smokeping is exactly the kind of background noise I'm talking about ping loss doesn't seem to make a huge difference between tiny and vewry tiny for results packetsize = 32 so i use that with fping brycec: mercutio : you guys are thinking too hard. admin's get SSH access to my ring box and I get SSH access to theirs. So everyone has access to everyone else's system. you can login to whichever one you want at a given time. If I see routing problems in Asia, I might pick a ring node that is located in that region, and then troubleshoot up_the_irons: That's exactly what I took away. yah rather than ping from every host 12:42:06 < brycec> So essentially this is a machine (VM) on your network that you allow other networks' admins access to (and in return, you get access to theirs) in order to troubleshoot the Internet? Is that essentially right? i was just thinking of things like dns and being able to check what caching is like etc brycec: yup that's right :) for some reason, the example of asia seems pretty fitting i dunno why asia has weird routing i think it's about routing policys most. but i'm in new zealand, which sometimes gets classed as asia pacific so some silly cdn's send traffic to asia rahter than the US which can end up going via the US to asia. actually japan seems pretty solid. it's singapore that's bad that i notice and from what i understnad india and pakistan and so forth are worse. wow, sold another dedi (and possibly even 1 more, will know later) they might be selling faster than VPSes this last week ;) up_the_irons: another one since my one? mercutio: yep the dedicated servers have only been going about 6 months right? mercutio: two actually, if the 2nd guy passes the OpenVPN/IPMI litmus test haha that's a test? :) shit i didn't know i was being tested. lol mercutio: a year; officially launched dedi's on 10/23/12, although several were sold before then oh wow time flies mercutio: yup you passed are there many colo servers? i prefer colo to dedicated when it's local but i imagine that doesnt' hold true for everyone mercutio: there's many colo'd, yes, 2 cabs; and i've been sold out of colo forever b/c those cabs are still full ahh and it's expensive to do another cabinet if it's not full up_the_irons: Any progress on the polo? well it's more that my cage got full, so i couldn't add another cab oh and expanding is expensive lol so does that mean you're running out of space in general? brycec: waiting for my tshirt to come and see if i like 'em are you at the primary or secondary dc? pretty damn near at this point damn i might grab a cab in another cage that i already have an x-conn to that a buddy of mine runs have you checked out the 10k sata disks btw? i wonder how they'd go for vm's dunno if they support tler or not http://www.newegg.com/Product/Product.aspx?Item=N82E16822236243 these Western Digital WD VelociRaptor WD1000DHTZ 1TB 10000 RPM 64MB Cache SATA 6.0Gb/s 3.5" Internal Hard Drive Bare Drive -- $229.99 they're meant to do 200mb/sec sequential pretty low power usage i think too, as 2.5" in an expansion tray which apparently isn't needed if in proper ventilated servers sounds like it should have tler, and similar errror rate per bits as re4 hey gizmoguy - ever had the opportunity to try stuff from New Belgium brewery in Ft Collins, CO? can't say I have they have some great stuff. #3 craft brewery in the U toured the brewery today, lots of fun + beer tasting awesome sounds like a good day I'm sitting here watching contractors fix my AC that died in my machine room over night.. yeah, just wish i had more time in colorado so i could hit odell and oskar blues much rather be drinking! oh damn, haha luckily I have 4 days off at the end of this week to fly and go drinking :D lose anything in the eq room? nah, not a complete failure right on. ive got another 7 days off. finally getting to burn some pto and spend some time with my wife. she's been so busy with grad school and work that i rarely see her just a failure in the cooling department it still circulates air that's good. cooling failures suck. just room temp air instead of chilled air it dies once a year... gotcha oh, that's great the company that has the contract seem incapable of fixing it better get yourself certified to work on hvac gear so you can fix it right :P I've got their old slow guy at the moment i'm sorta hoping they escalate to someone who moves a bit faster cause I want to go home at the end of the day :P haha true so far this guy turned up I explained problem and then he pulled out his ladder and then I heard lots of clunking and smashing and some swearing and I kinda walked away after that ah crap brb http://i.imgur.com/ngaV7Fp.png what is this guy doing?! hah Ok so he turned the AC unit off and then walked away apparently without telling me "fixed!" .... so I'm attempting to cool with my backup cooling system which involves opening some doors :P up til recently, my backup cooling was opening up the ceiling to let the hot air escape oh holy crap i'm in like a hundred channels on freenode i don't remember joining half of them i'm in exactly one channel on freenode: this one Good, the most important one! I'm in about 20 or so, most of them fairly quiet. im just not a big fan of freenode oh counting just FN? I think just 4 of those are FN channels 1 windows used (0 vertically / 0 horizontally split). 56 (of which 0 merged) buffers open: 1 core, 12 irc servers, 2 perl, 41 irc channels used to hang out in #pfsense and #m0n0wall years ago 56 channels? geez i dont think i could find that many channels that id care enough about to even idle in m0unds: you picked the right channel :)