[00:02] @wa date 51 days from now [00:02] current time+51 days;3:02:36 am EDT -> Friday, August 1, 2014 [01:02] Reading reviews for Mikrotik Routerboard RB2011UAS-2HND-IN 1 SFP port plus 10 port Ethernet on Amazon, and came across: This is advertised as having 5 gbps ports, and 5 100mbps ports. What they fail to mention is that the 5 "gbps" ports all share the 1000mbps - so you really only get 1000/5 = 200mbps per port. [01:02] Is that true? the Mikrotik oversubscribes the gigE LAN ports? [01:05] mnathani: http://routerboard.com/RB2011UiAS-2HnD-IN [01:06] scroll to "performance" [01:06] they might be refering to "routed" speed rather than switched speed? [01:16] kpps < is that x1000s packets per second? [01:16] yup [01:17] I wonder how it compares to regular Linksys routers that have 4 port gigabit switches built in [01:18] 128 MB vs 32 MB would certainly make a difference [01:18] RAM [01:22] ram quantity doesn't make much difference to routing performance [01:22] cpu then/ [01:23] it's the same cpu model as tp-link 4300 [01:23] but 40mhz difference in clock rate [01:23] i don't think it'll make much difference [01:23] nearly all the routers are mips based [01:24] i imagine once adsl becomes less common more people will move to arm [01:24] RouterOS better at moving packets than tp-link os/firmware [01:24] you can get mips cores with built in adsl [01:24] mnathani: doubt it [01:24] they're both linux [01:26] i had a d-link 504t adsl modem that's cpu couldnt' keep up with adsl2+ speeds [01:26] but it had 16mb of ram and could run openwrt with adsl support [01:26] but other than that i've never hit real cpu bottlenecks. [01:27] even a pentium 75 can route 100 megabit of large pcaket sizes [02:41] *** toddf_ has quit IRC (Ping timeout: 252 seconds) [02:42] *** toddf has joined #arpnetworks [02:42] *** ChanServ sets mode: +o toddf [02:51] too early to be at work. bleck. [02:52] what's the time [02:54] left home at 0315 [02:54] hahaha [02:55] :/ [02:55] yeah, doing our big shutdown event this morning. woohoo. [02:56] fun fun [02:56] yup, just firing off all our backups and stuff real quick [02:57] nah [02:57] shutdown -h now [02:57] on everything [02:57] or just pull the plug [02:57] it'll be all g [02:57] haha [02:57] was gonna say i only have a single linux box [02:57] shutdown /h /t now [02:57] or something? [02:57] something like that [02:57] dunno not a windows d00d [04:19] *** tabthorpe has quit IRC (Ping timeout: 265 seconds) [04:19] *** medum has quit IRC (Ping timeout: 265 seconds) [04:20] *** medum has joined #arpnetworks [04:21] haha [04:25] *** tabthorpe has joined #arpnetworks [06:16] *** gizmoguy has quit IRC (Ping timeout: 245 seconds) [06:17] *** gizmoguy has joined #arpnetworks [07:32] *** grepidemic has joined #arpnetworks [07:36] *** toddf has quit IRC (Ping timeout: 264 seconds) [07:37] *** toddf has joined #arpnetworks [07:37] *** ChanServ sets mode: +o toddf [07:41] *** toddf has quit IRC (Ping timeout: 260 seconds) [07:42] *** toddf has joined #arpnetworks [07:42] *** ChanServ sets mode: +o toddf [07:47] *** toddf has quit IRC (Remote host closed the connection) [07:48] *** toddf has joined #arpnetworks [07:48] *** ChanServ sets mode: +o toddf [07:49] that was fun [07:50] https://twitter.com/VICE/status/476737539081261056 [07:50] TWITTER: The Strange Swiss Custom of Dressing Up as a Bush and Throwing Women in Wells http://bit.ly/1hOEw5C http://t.co/UmVEWVWzcW (Wed Jun 11 14:47:41 +0000 2014, retweeted 28 times) [07:52] *** toddf has quit IRC (Ping timeout: 260 seconds) [07:53] *** toddf has joined #arpnetworks [07:53] *** ChanServ sets mode: +o toddf [08:25] *** toddf has quit IRC (Ping timeout: 265 seconds) [08:26] *** toddf has joined #arpnetworks [08:26] *** ChanServ sets mode: +o toddf [08:31] *** toddf has quit IRC (Remote host closed the connection) [08:31] *** toddf has joined #arpnetworks [08:31] *** ChanServ sets mode: +o toddf [08:56] *** m0unds_ has quit IRC (Ping timeout: 245 seconds) [09:04] *** m0unds_ has joined #arpnetworks [11:00] http://www.moogmusic.com/news/introducing-werkstatt-01 i want one [11:38] *** staticsafe|2 has quit IRC (Read error: Connection reset by peer) [11:43] sounds german [11:48] yeah, "workshop" [11:48] it's a build it yourself mini-moog project synth [11:48] they package the chassis, pcb and stuff, and you assemble it [13:13] cool [13:39] *** koan has quit IRC (*.net *.split) [13:39] *** hazardous has quit IRC (*.net *.split) [13:39] *** qbit has quit IRC (*.net *.split) [13:40] *** koan has joined #arpnetworks [13:40] *** qbit has joined #arpnetworks [13:40] *** koan has quit IRC (Changing host) [13:40] *** koan has joined #arpnetworks [13:40] *** hazardous has joined #arpnetworks [14:10] *** z310 has quit IRC (Excess Flood) [14:12] *** z310 has joined #arpnetworks [14:58] *** carvite has quit IRC (Remote host closed the connection) [15:10] *** carvite has joined #arpnetworks [16:09] up_the_irons: just wondering if you'd seen my support ticket regarding the NTT issues [16:10] I put some mtrs in there, but I can make more/different ones if it would help [17:58] *** wacker has joined #arpnetworks [18:02] Hi all. I have a fun idea. If back01.cust.arpnetworks.com had an AAAA record, customers could add the /128 to their routingtables on their VPS's pointing the Link-local for the backup server over the dedicated interface and magic would happen! [18:32] Of course i meant backup01.cust.arpnetworks.com. How I wich I could type. [19:32] 10. comstock.cable.packetflow.ca 80.0% 35 24304 24186 23986 24365 152.6 [19:32] is that really 24 seconds ? [19:49] Or... those users could just add it to their /etc/hosts :p [19:50] fe80::21b:21ff:fe93:3e08 arp_backup [19:50] (though I just use the IP in my commands) [19:51] (IP%interface) [20:04] blah [20:05] *** toddf has quit IRC (Ping timeout: 265 seconds) [20:05] *** toddf has joined #arpnetworks [20:05] *** ChanServ sets mode: +o toddf [20:05] oh neat, there was a trailer for elite: dangerous shown at e3 [20:28] [20:28] [20:28] [20:28] [20:28] [20:28] oops [20:29] * mnathani slaps brycec around with a wet trout for spamming the channel [20:29] with blank lines too [20:29] "empty spam" [20:29] heh [20:30] (Supper plate was slightly resting upon the spacebar) [20:33] @date eastern [20:33] I'm strtotime, not a mind-reader and not an encyclopedia. Give me a date I can understand. (http://php.net/manual/en/datetime.formats.php) [20:33] @date EST [20:33] 2 hours, 0 seconds ago. [Interpreted date: Wed, 11 Jun 2014 18:33:21 -0700] [20:33] @date EDT [20:33] 3 hours, 0 seconds ago. [Interpreted date: Wed, 11 Jun 2014 17:33:30 -0700] [20:33] @date [20:33] Hey, that's right now!! [Interpreted date: Wed, 11 Jun 2014 20:33:49 -0700] [20:33] @date NZST [20:33] 19 hours, 0 seconds ago. [Interpreted date: Wed, 11 Jun 2014 01:33:56 -0700] [20:34] is it really 20:33 wherever BryceBot it located? [20:34] I guess that PST [20:34] @date PST [20:34] 1 hour, 0 seconds to go. [Interpreted date: Wed, 11 Jun 2014 21:34:38 -0700] [20:34] Yes, BryceBot is located in Los Angeles, at ARP Networks [20:34] @date PDT [20:34] Hey, that's right now!! [Interpreted date: Wed, 11 Jun 2014 20:34:45 -0700] [20:35] @wa time in EST [20:35] @wa time in Toronto [20:35] Error fetching URI. [20:36] Sorry, I couldn't reach the backend API. [20:36] brycec: Putting the link-local in /etc/hosts doesn't seem to work on Linux. Of course, you could put it in all of your scripts, as you say, but I think my solution should work as well for those of us who want to use it. [20:36] @wa 2+2 [20:36] Error fetching URI. [20:36] bot broke [20:36] mnathani: WA broke [20:37] wacker: Works for me, but doesn't include the %interface of course. [20:37] Specifically, WA is timing out for BryceBot [20:38] I blame Level3 [20:40] @wa date 20 days from now [20:40] current time+20 days;11:40:17 pm EDT -> Tuesday, July 1, 2014;19 minutes 43 seconds [20:40] @wa time in Toronto [20:40] current time in Toronto, Ontario, Canada;11:40:47 pm EDT -> Wednesday, June 11, 2014;19 minutes 13 seconds;3:40:47 am GMT -> Thursday June 12, 2014 [20:41] @wa time in brazil [20:41] current time in Brazil;12:41:24 am BRT -> Thursday, June 12, 2014;+1 hour;Brasília, , 12:41:24 am BRT, Thursday, June 12 New York City, New York, , 11:41:24 pm EDT, Wednesday, June 11;3:41:24 am GMT -> Thursday June 12, 2014 [20:55] *** d^_^b has quit IRC (Ping timeout: 252 seconds) [20:56] *** d^_^b has joined #arpnetworks [20:56] brycec: With %eth1, I get "unknown host" without it, I get "invalid arguement" when I ping6. So, I can leave it off, and do ping6 -I eth1 backup01. That works for ping, but not so much for ssh and friends, the commands I realy want to use. [20:56] *** d^_^b has quit IRC (Changing host) [20:56] *** d^_^b has joined #arpnetworks [20:57] Reply from 192.168.2.254: bytes=32 time=241ms TTL=64 [20:57] every once in a while my home router forgets how to route and does ^ [21:00] wacker: A AAAA record will work the same as a /etc/hosts entry. You'd still have to add the /128 route [21:07] huh, i guess windows update pulls down via ipv6 if available [21:08] So true, but I'd expect the AAAA record to be a global, not a link-local. [21:39] WA? [21:51] Washington, Wolfram Alpha, that's all I can think of [22:04] acf_: yeah, i got it, just been slow on tickets for the last couple days [22:04] wacker: hmm, but backup01.cust will have a different address for each link-local (each customer, that is) [22:05] oh wait... [22:05] no... [22:05] * up_the_irons puts his brain back in [22:06] if you give it an AAAA record, it could potentially interfere with folks outside ARP trying to back up stuff [22:07] mnathani: ah good point [22:08] even though i did not mean for the backup server to be accessible outside... it seems that a lot of people have a valid use for an outside interface, so i keep it [22:09] I like that I can use it to backup servers in different regions across the US over to a secure West Coast Facility (ARP) [22:12] see :) [22:13] now, technically, you could be using a lot of bandwidth and there's no way for me to charge for that, since the outside interface is public and doesn't belong to any single customer VLAN. oh well, good thing i have more bandwidth than I use. [22:14] so kvr15 is on the chopping block tonight [22:14] you could do something on the OS level that tracks scp / rsync per login [22:14] 3rd server this week, i feel so productive [22:14] yeah, but no time to research / implement that [22:14] * brycec quietly slurps up bandwidth, a couple MB at a time [22:15] up_the_irons: have you announced any ip blocks to level3 as yet? [22:15] not yet [22:15] brycec: hah [22:16] busy with other stuff this week (mostly upgrades).. man it never ends [22:16] Go up_the_irons ! [22:16] :D [22:17] do the kvr hosts have an expected lifetime, before they are eventually replaced? [22:18] a VPS on kvr15: 01:17:51 up 755 days, 20:32, 1 user, load average: 0.00, 0.00, 0.00 [22:19] nice [22:20] its a VPS I manage on behalf of a customer [22:31] mnathani: no set lifetime [22:32] or expected. [22:32] i've had 1 fail, out of 30; it failed within the first year i think, so i think the MB was a dud [22:32] what model machines do you use? [22:32] it was kvr25 [22:32] acf_: supermicro [22:33] cool. they sound nice :D [22:33] the newer onces are AMD, correct? [22:33] s/onces/ones [22:33] the newer ones are AMD, correct? [22:33] acf_: :) [22:33] mnathani: yes, newer ones are AMD [22:33] are they the same as the arp metal ones? [22:33] no [22:33] arp metal are Intel E3 blades [22:34] they don't have near enough HD slots to be a kvm host [22:34] ah, right [22:34] turns out 9/10 metal customers never order more than 2 HDs, so it works [22:34] but for a kvm host, you need the 8x bays [22:35] hot swappable ofcourse? [22:36] oh for absolute sure [22:36] can't take down a whole host just cuz an HD fails (and they die like flies) [22:38] yeah. those are some awesome uptimes [22:38] like 1500 days? [22:38] up_the_irons: do you have your own UPS or do you rely on those provided by the facility [22:39] acf_: yup, kvr06 and kvr07 are over 1500 days [22:40] @wa 1500 days in years [22:40] convert 1500 days to years;4.11 years;49.32 months;214.3 weeks;36000 hours;2.16×10^6 minutes;1.296×10^8 seconds;~~ 0.35 × orbital period of Jupiter (~~ 12 Julian years );~~ 0.78 × half-life of cobalt-60 ( 1.6635×10^8 s );~~ 1.6 × half-life of sodium-22 ( 8.2108×10^7 s );time;period [22:40] mnathani: provided by data center (N+1 UPS, redundant PS fed from A/B circuits on separate elec panels) [22:40] over 4 years. WOW [22:41] mnathani: i really put a lot of thought into the design 5 years ago ;) [22:41] yeah that's insane [22:41] check this out: [22:41] s1.lax#sh ver [22:41] Cisco IOS Software, Catalyst 4500 L3 Switch Software (cat4500-ENTSERVICESK9-M), Version 12.2(31)SG, RELEASE SOFTWARE (fc2) [22:41] ... [22:41] s1.lax uptime is 5 years, 16 weeks, 2 days, 10 hours, 1 minute [22:41] :) [22:41] it was s7 that was crashing then? [22:42] yeah [22:42] how many ports on that s1 ? [22:42] 48 FastE, 8 GigE [22:43] well, thats cisco for you: solid as rock [22:44] so, you just have a bunch of cabinets adjacent to each other [22:44] with switches at the top [22:44] and you run ethernet back to the routers? [22:45] mnathani: yep [22:45] how many nics in the kvr hosts, and are atleast 2 setup as trunk ports to carry the VPS VLANs ? Second one for redundancy + performance? [22:45] acf_: i run ethernet back to core. i'd rather buy longer cables than manage more switches. [22:46] mnathani: 4 NICs. 1 - s1.lax, 2 - s6.lax (internal use), 3 - s7.lax (redundant for s1.lax), 4 - unused [22:47] hmm. so is everything routed through s7 pretty much? [22:47] like that's where all the bgp is? [22:48] bqp sessions are on BIRD? [22:48] acf_: most traffic goes out s1.lax. BGP is on both and both carry inbound traffic. [22:48] s1 - GTT, s7 - NTT, Level 3 [22:48] mnathani: only for peering [22:49] I'm guessing that's 10.10.10.6... [22:49] the peering box [22:49] IPv6 is on a software router as well? [22:49] acf_: yes [22:49] mnathani: yes [22:50] IPv6 is an OpenBSD box running OpenBGPD [22:50] oh cool [22:51] just out of curiosity, why use an rfc1918 address for the peering box? [22:51] isnt it due to ip addressing of the internet exchange coresite/any2 [22:53] mnathani How would giving the backup server an AAAA record hurt anything? Most of the systems over there are dual stacked, or so it seems. [22:53] it would only work from within ARP [22:54] say someone from outside ARP tries to connect on IPv6, it would try and it would fail as that is not a global IPv6 address [22:54] The v4 is accesible globally, why not the IPv6. As I said, I'm looking for a global like 2607:f2f8:. [22:55] * up_the_irons must jet, Starbucks is closing [22:55] IPv6 at ARP is run on a software router. I dont think up_the_irons would want backup traffic using up IPv6 bandwidth / cpu cycles [22:56] Starbucks closing at 11pm... madness :( [22:56] Don't miss the coffee, Garry, you're probably up for the night in any case. [23:00] http://paste.unixcube.org/k/7f55e [23:00] funny you can see the rfc1918 addresses from both sides [23:00] I would have thought some ingress filters might block packets from them [23:21] acf_: rfc1918 for peering.. well, why not? it's not going over public internet and so why waste a /30 [23:22] mnathani: yeah, public backup traffic over ipv6 software router would not be good [23:22] brycec: love my local Starbucks. most don't stay open up to 11pm. [23:23] wacker: i actually get tea these days at night, or else i'm up until 7am [23:23] Lucky you, up_the_irons [23:23] (My closest Starbucks closes at 8:30) [23:32] Is the upgrade to 12.04 done remotely over ipmi or do you actually go to the datacenter [23:40] brycec: man what a crappy time [23:40] mnathani: remotely over ipmi [23:48] for some reason Starbucks always seemed like a 24/7 kind of place to me [23:48] not a coffee drinker, so I wouldn't know [23:48] I never understood why a coffee place wouldn't stay open late [23:58] i know, right