[00:08] *** Webhostbudd has quit IRC (Ping timeout: 245 seconds) [00:44] *** HighJinx has joined #arpnetworks [08:25] *** dj_goku has joined #arpnetworks [08:25] *** dj_goku has quit IRC (Changing host) [08:25] *** dj_goku has joined #arpnetworks [08:39] *** dj_goku has quit IRC (Ping timeout: 245 seconds) [11:33] *** mhoran has quit IRC (Quit: WeeChat 0.3.9) [12:01] *** Webhostbudd has joined #arpnetworks [12:09] *** mhoran has joined #arpnetworks [12:09] *** ChanServ sets mode: +o mhoran [12:10] whoa, lots of scrollback [15:16] *** Mexicainvexed has quit IRC (Ping timeout: 245 seconds) [15:21] *** dzup has joined #arpnetworks [15:28] *** rith has quit IRC (Quit: EOF) [15:53] You're welcome! (not that I had anything to do with it..) [15:53] * RandalSchwartz mutters "happy new year" [15:54] * brycec yawns at 5pm [15:54] yeah - only 4pm here [15:55] but it's over the top for some folks [16:03] *** Mexicainvexed has joined #arpnetworks [16:03] it's been the new year for ages [16:03] over 13 hours :/ [16:03] UTC is the only relevant time zone :P [16:03] *** dzup has quit IRC (Ping timeout: 245 seconds) [16:03] you mean gmt? [16:03] UTC [16:04] isn't utc just a reference for other time zones [16:04] UTC was used beginning in the mid-twentieth century but became the official standard of world time on January 1, 1972. [16:05] oh i didn't know it being an official standard [16:05] root@hellhound ~ # date -u [16:05] Tue Jan 1 00:04:55 UTC 2013 [16:05] :) [16:06] # date -u [16:06] Tue Jan 1 00:06:02 UTC 2013 [16:06] your click is behind [16:06] oh actually... i think they're about the same [16:06] i think my irc client is ahead [16:06] all my servers are synced via NTP [16:07] most of mine arc [16:07] it seems the box i irc on isn't though [16:07] Tue Jan 1 00:07:35 UTC 2013 [16:07] my irc client +37 seconds to you [16:07] Tue Jan 1 00:07:43 UTC 2013 [16:07] running ntp, so that should be close [16:08] same [16:08] 2 seconds diff between you two [16:08] from my perspective [16:08] can't seem to run ntp on any box that has to be PCI compliant though [16:08] there of course can be latency between irc servers/connections [16:08] so instead I run ntpdate once a day. [16:08] dumb solution [16:08] mercutio: yes indeed [16:08] RandalSchwartz: wtf [16:09] yeah - they keep flagging ntpd on freebsd as "bad" [16:09] so the only solution is to not run it [16:09] openntpd? [16:09] or that huge bloated one? [16:09] they just flat the open tcp port [16:09] flag [16:09] oh [16:09] so i've had to turn that off [16:09] idiots [16:09] does ntp use tcp? [16:09] UDP [16:10] oh i thought it may have a fallback like dns [16:10] but they apparently tweak it, and it comes back, so they say "you're bad" [16:10] idiots [16:10] PCI compliance is a joke [16:10] welcome to politics :) [16:10] but my $client has to pass, or not take credit cards from $bank [16:10] heh [16:11] some of the stuff makes sense [16:11] but there's a huge pile of WTFery mixed in [16:12] does anyone use the calculator key on keyboards? [16:12] I mean, yeah, not storing CVVs or unencrypted PANs, all good. [16:12] what's "The calculator key" [16:12] the picture of a calculator [16:12] * RandalSchwartz looks at his keyboard [16:12] I have none of those [16:13] it's on most modern keyboards [16:13] I have a very modern keyboard [16:13] macbook pro 17-inch [16:13] miune is where print screen would usually be :( [16:13] late 2011 or so [16:13] oh, laptops don't count [16:13] sure they o [16:14] they don't even have a number pad! [16:14] sure they do [16:14] it's just buried in the middle of the KB [16:15] my media controls are like that [16:15] .. https://discussions.apple.com/thread/3332788?start=0&tstart=0 [16:17] O_o [16:17] * RandalSchwartz wanders off to drive home [17:09] *** Mexicainvexed has quit IRC (Ping timeout: 240 seconds) [19:50] There are keyboards attached to computers that aren't laptops? [19:51] what? [19:51] you mean all in one? [19:51] like mac ? [19:51] imac [19:51] and amiga [19:52] Hah, I dunno. I've only used laptops and servers for the past five or 10 years. [19:52] is macbook pro the g5 one? [19:52] really? [19:52] g5? [19:52] i have a laptop, i use it a few times a year [19:52] g5 cpu [19:52] my macbook pro is an i7 [19:52] oh that's powerbook [19:53] ... http://en.wikipedia.org/wiki/MacBook_Pro [19:53] oh they never actually did it [19:53] i7s in laptops are usually i5s [19:53] renamed as i7s [19:53] no [19:53] and i5s i3s [19:53] real i7 [19:54] the previous one I had was an i5 [19:54] and I feel the difference [19:54] Intel Core i7-2620M @ 2.70GHz Average CPU Mark [19:54] Description: Socket: BGA1023, Clockspeed: 2.7 GHz, Turbo Speed: 3.4 GHz, No of Cores: 2 (2 logical cores per physical), Max TDP: 35 W [19:54] dualcore i7? :) [19:54] well that's cos i5 is i3 [19:55] I have 8 threads [19:55] really [19:55] wow [19:55] yes [19:55] which cpu? [19:55] this was early 2011 cpu [19:55] ... 2.4 GHz (2760QM) quad-core Intel Core i7 Sandy Bridge with 6 MB on-chip L3 cache [19:55] yes [19:56] oh cool [19:56] "MacBookPro8,3" [19:56] yeh it sucks how intel do that [19:57] oh [19:57] i7 normally has 8mb cache or more i think [19:57] actually - it says I have 2.5 ghz [19:57] so it's even better than that [19:57] oh heh [19:57] http://ark.intel.com/products/65719/ [19:57] and yes 8 MB L3 [19:57] desktop i7 is 3.5 ghz and 8mb cache [19:58] oh you have 8mb cache cool [19:58] this is according to system update [19:58] do you have hd4000 video? [19:58] my laptop is damn slow [19:58] AMD Radeon HD 6770M [19:58] and Intel HD Graphics 3000 [19:59] core2duo 2.4 ghz or something [19:59] with 4gb ram [19:59] and 8gb ram? [19:59] and the retina screen thingy? [19:59] no - looks like 1024MB for the first, and 512 for the second [19:59] oh [20:00] no retina screen [20:00] this wasn't from 2012 :) [20:00] i7 with 1.5gb ram? [20:00] ahh [20:00] so it's 1680x1050? [20:00] just 1920x1200 [20:00] oh cool [20:00] that's better than my laptop [20:00] i only have 1680x1050 :( [20:00] i like how apple are pushign resolutions up [20:01] *** Lefty has quit IRC (Remote host closed the connection) [20:01] at least it has dedicated video card :) [20:02] the new macs look ok [20:02] err mac laptops [20:02] oh [20:02] but they only take 8gb ram :( [20:03] Optional 16 GB RAM configuration available at time of purchase only [20:03] oh [20:03] they do take more [20:03] yeah I have mine baxed out [20:03] you jhust have to order it at time [20:03] maxxed out [20:03] at 8 [20:03] 16gb is standard for reasonably heavy users now [20:03] if buying new [20:03] I've never gotten up to all of that. [20:03] but laptops often only have two slots [20:03] I guess if you're running vm's or something [20:03] yeh that's cos you're using a laptop [20:03] 8gb sticks are cheap for desktops now [20:03] you can have 32gig for prob USD $200? [20:03] *** Lefty has joined #arpnetworks [20:04] it makes a big diff to everything really. well, 16gb does compared to 6gb [20:04] that's why i upgraded from ddr2 to ddr3 [20:04] but that means you have to replace cpu... motherboard.. etc to [20:04] too [20:05] i wonder how much these mac laptops cost [20:05] $3500 [20:05] ouch [20:05] very price appropriate, given hardware features [20:05] no more "mac penalty" [20:06] desktops are a lot cheaper :) [20:06] sure, and have fewer things [20:06] more things you mean? [20:06] Different purposes. [20:06] more expandability bigger monitors, bigger ssds etc [20:06] but if you price it out, feature for feature, mac hardware is very competitive [20:06] mike: depends if being used as a desktop too [20:06] yeh you can't get good screens on normal laptops [20:06] at least i haven't seen one recently [20:07] the trouble is that people tend to compare cheap crappy desktops against macs [20:07] macbook air is not too expensive [20:07] and yeah, so cheap crappy desktops are cheaper. duh. [20:07] cheap crappy? [20:07] i wonder how much my desktop is worth [20:08] it'd probably be $2000 or something [20:08] to replace every bit [20:08] probably slightly more than that [20:08] oh wow [20:08] it's easy to forget how much computers cost when you add things over time [20:09] like i got a good video card, so i can do high res... [20:09] which it does [20:09] but good video cards are still expensive [20:09] and even then i'm only sitting at 70 fps [20:09] It's still slightly surprising to me how many different things people use computers for. [20:09] I use mine to heat my room [20:09] facebook? [20:10] google [20:10] Like, how I'd never do someting that involves high res. [20:10] i realised the toher day, i never use my laptop without the internet [20:10] like if i didn't have internet for some reason, i'd think of my laptop as being useless [20:10] mike: reading pdfs? [20:10] I use my phone for PDFs. [20:11] is your phone high res? [20:11] Doubtful. [20:11] i have to scroll like mad on my phone to read pdfs. [20:11] It's just more convenient on the metro. [20:11] my tablet is slightly better [20:11] metro? [20:11] Tunnelbana. [20:11] train? [20:11] Sure. [20:11] what are trains like for internet these days [20:12] yay, 1.5G of RAM makes for a happy VPS [20:12] heh my vps went from 256 to 768mb [20:12] but i'd prob be fine with 256 still [20:12] I went from 768 to 1.5 [20:12] but I'm running a redis server, a mysql server, and a bunch of tmux [20:12] The subway offers no wifi here. The only other train I take is the airport shuttle, and that does offer free wifi. [20:12] whereas my desktop swapped heavily on 6gb [20:13] all of those like moar RAM [20:13] that said, i don't use chrome on my vps [20:13] lefty: tmux isn't memory hungry [20:13] heh, I get cranky if my desktop has any less than 16 these days [20:13] but yeah the other two could be [20:13] Web browsers take crazy amounts of RAM. I blame animated GIFs. [20:13] tmux consistently is the highest ram consumer on my box [20:14] i'm actually only using 8.4 gig on my desktop atm [20:14] Is the tmux RAM measurement being reported properly? That is, is it instead reporting all the RAM of its child processes? [20:14] one chrome window on it's own is using 450mb [20:14] mike-burns: entirely possible [20:14] mike: it doens't report child processes normally? [20:15] at least not under openbsd or linux [20:15] I didn't look into it that deeply, but the box was usually a few hundred megs into swap most of the time [20:15] 14600 root 2 0 668K 1332K sleep kqread 0:00 0.00% tmux [20:15] PID USERNAME PRI NICE SIZE RES STATE WAIT TIME CPU COMMAND [20:15] that doesn't seem memory hungry to me [20:15] 13146 _mysql 2 0 40M 15M idle select 0:30 0.00% mysqld [20:15] mysql takes up a bit more [20:15] before I rebooted, it was reporting ~300M [20:16] that's light mysql usage though [20:16] wow [20:16] it's probably leaking then [20:16] mysql? friends don't let friends use mysql. [20:16] actually i'm using tmux on my irc [20:16] times four or so users [20:16] and it's been running ages [20:16] ben 16396 0.1 0.3 4572 2440 ? Ss 2012 7:58 tmux -2 [20:16] that's linux instead that time [20:16] apparently, I'm not your friend yet. :) [20:16] 2.4 mb resident? [20:17] I'm extremely pleased with the trend of ditching MySQL for Postgres. [20:17] i wonder how long it's been running [20:17] if you insist on mysql compatible, use mariadb [20:17] but otherwise, postgresql, yes [20:17] in either case, please move away from the oracle frankenserver [20:17] i was thinking the other day, mysql is terribly slow [20:18] surely their query optimisation could be a lot faster on modern computers? [20:18] err could lead to much faster performance [20:18] like current performance compared to ideal performance [20:18] mysql sucks on modern hardware [20:18] i imagine there's lots of complications of such [20:18] but like say you have 16gb ram [20:18] well, it sucked on old hardware too, it's just now modern hardware tends to have lots of cores and RAM [20:18] and 2gb database [20:18] mariadb is drop-in-completely-replacement for mysql [20:18] monty promises such [20:19] it doesn't really matter if you use 8gb of ram [20:19] and do things to reduce cpu bottlenecking etc [20:19] and it's better/stronger/faster [20:19] so absolutely no reason to run mysql. evar. [20:19] randal: mysql comes bundled with operating systems though? [20:20] apt-get is hard to type? [20:20] Which OS bundles mysql? [20:20] does ubntu have maria now? [20:20] mike: ubuntu? [20:20] Odd. [20:20] ... https://downloads.mariadb.org/mariadb/repositories/ [20:20] ubuntu, centos, openbsd, etc all have mysql support easily [20:20] serious, google [20:20] please don't make me google for you [20:21] Is it useful for browsing the Web and, I dunno, whatever else Ubuntu users do? [20:21] it doesn't have my ubuntu version [20:21] which version of ubuntu? [20:21] hardy to precise [20:21] yeah i use quantal [20:22] yeah, be weird. get source then [20:22] no mariadb in my repos for 12.04 on arm [20:22] lefty: you have to get it from external repository [20:22] nor x86 [20:22] but that's why people use mysql [20:22] well, I still have to get mysql from an external repo [20:22] when it doesn't matter it's easier to use mysql [20:22] at least on our centos boxes [20:23] when it does matter then people are like "why should you use mariadb" [20:23] there was mariadb and another one that i was thinking about migrating to [20:23] Postgres. [20:23] nah [20:23] my mysql usage at home is just a toy database [20:23] i've used postgresql in the past [20:23] at work, though, that's a different story... we do some heavy mysql scariness [20:24] but lots of people are more familiar with mysql [20:24] slaves of slaves of slaves [20:24] lefty: with redis? [20:24] in fact, the only alerts I've got for this on-call was mysql replication breaking [20:24] couchbase 2.0 looks pretty damn cool [20:24] mercutio: we use a lot of redis, too, and a lot of mongo [20:24] why does mysql insisit on using disk for queries anyway [20:24] materialized views [20:24] hmm wonder what one i looked at [20:25] percona [20:25] I may be deploying it for $client [20:25] i thik that was it [20:25] looking at wikipedia [20:27] so is mariadb a lot faster than mysql in common usage patterns? [20:27] monty claims so [20:27] but like 50% faster? [20:27] I. Don't. Know. [20:27] ahh [20:34] looks like most of the improvments only happen with 16+ cpu cores [20:35] and still 20% from worst to best [20:37] heh, that'd help our workload [20:37] the biggest problem with mysql replication is that slaves are entirely singlethreaded [20:37] so we can have giant boxes with lots of cores etc, and they don't get used at all [20:38] and replication still lags [20:38] I can't imagine maria is any better in that respect, so we'll probably stick with mysql [20:39] yeh i'm sure it'd help heavy heavy workloads [20:39] we have mysql boxes that are 16core [20:39] hmm, i didn't find anything about replication [20:39] can you do replication int he app? [20:39] the [20:39] yes, but nobody wants to touch it [20:39] it's very, very legacy [20:40] i was mostly wanting stupid web applications to go faster :/ [20:40] it's slowly being replaced by modern systems that use better storage technologies [20:40] though by "better" I'm using it in the loosest sense of the word [20:40] yeh [20:40] mongodb didn't get production-ready until 2.2 [20:40] anything before that was fucked hard [20:41] fwiw in these benchmarks percona seems better than mariadb [20:41] at 2, 8, 16 cores [20:41] weirdly the difference is less at 32 cores [20:41] then percona is way way faster at 64 core [20:42] 4200 versus 2600 tps [20:42] neat [20:42] * Lefty found out the other day that redis can't handle more than 32G of RAM allocated to it [20:42] heh [20:42] you must have a lot of data [20:42] we generate a lot of billing data [20:43] a while back i read about memcached not scaling [20:43] and the funny thing is that the billing group always says "oh, we'll only need a couple of gig of RAM", then they fire of a few million subscriptions and wonder why redis fell over [20:43] off, too [20:43] due to memory being way worse with random reads than linear [20:43] i think you're meant to use low latency memory if you want random queries [20:44] anything, really [20:44] I mean, I can't think of many workloads that consist of sequentially reading large amounts of data into RAM, then reading it right back out again [20:44] you don't know my clients. :) [20:44] oh i think it was the facebook discussion [20:45] I can see reading a lot of data, then accessing it randomly [20:45] lefty: blogs? [20:45] * Lefty shrugs [20:45] i also read at one point that l3 cache isn't a lot faster than memory [20:45] but has lower latency.. [20:46] i assume cos cache is sram [20:46] it's like when memory went to sdram the peformance improvement wasn't that great [20:46] because the cpus couldn't make that good use for the burst memory speed [20:47] although that changed over time [20:48] curiously i was reading just now that facebook got better performance from shifting from tcp/ip to udp/ip. [20:48] i wonder how much better performance could be with infiniband [21:30] infiniband is gimicky from what i heard [21:30] although definitely better than gige [21:56] what's gimicky about it? [21:56] it's significantly lower latency than 10 gigabit ethernet [21:57] it's also cheaper [21:57] especially if you want to go over 10 gigabit throughput [21:57] then it's got things like rdma to reduce overhead [21:58] whereas infiniband guarantees in order packet delivery [23:37] well, i've heard 10gige performs better than 20gig infinibamd [23:37] however, i don't know the network layourt [23:51] infiniband native? [23:51] or doing ethernet emulation?