| --- | Log | opened Thu Mar 13 00:00:02 2008 |
| --- | Day | changed Thu Mar 13 2008 |
| 00:00 | <JasonF> | I've suggested a quadrillion times that linode peoples provide a portage tree over NFS |
| 00:00 | <JasonF> | I might would still be using gentoo if they did |
| 00:01 | <guinea-pig> | mmm, i should play with gentoo again |
| 00:01 | <guinea-pig> | haven't played in a couple years now |
| 00:02 | <Talman> | Do they pay for bandwidth between boxes in the racks? |
| 00:03 | <JasonF> | nah, that's local |
| 00:04 | <martinisoft> | Yeah the prospect of compiling everything is making me not select Gentoo |
| 00:04 | -!- | clanehin [~lane@cpe-069-134-066-130.nc.res.rr.com] has quit [Ping timeout: 480 seconds] |
| 00:04 | <martinisoft> | so now its a tie between Slackware and Ubuntu |
| 00:05 | <Talman> | mwalling loves him some slack. |
| 00:05 | <booja> | martinisoft: I use slack, I know mwalling does too |
| 00:05 | <martinisoft> | I'll probably go with Ubuntu because its very newbie friendly, but Slackware's hardline approach to sysadmining is very appealing |
| 00:06 | <booja> | just be aware that the linode slack12 image is missing quite a log of stuff |
| 00:06 | <martinisoft> | yeah its *very* small, which seems nice |
| 00:06 | <booja> | first thing is to install slackpkg and point it at a mirror :) |
| 00:06 | <martinisoft> | I'd rather start with a base, clean image |
| 00:07 | <booja> | there's clean, and there's unfunctional |
| 00:08 | <Talman> | There's clean and there's "you get sh." |
| 00:09 | <booja> | if you're lucky, bash'ed :P |
| 00:12 | <martinisoft> | by clean I mean base packages, shell and compiling tools, nothing else |
| 00:12 | -!- | john [~d8600bfa@webuser.linode.com] has quit [Remote host closed the connection] |
| 00:12 | <martinisoft> | enough to start a custom web stack |
| 00:12 | <martinisoft> | I want nginx and mysql |
| 00:12 | <Talman> | Do you have compiling tools on their slack gimmick, booja? :) |
| 00:13 | <martinisoft> | mostly because I've come to hate Apache |
| 00:13 | <Talman> | Or are you getting to import those. |
| 00:13 | * | Talman has no problems with apache, but you are doing rails. |
| 00:13 | <martinisoft> | I come from the Solaris/BSD world, so Linux is... yeah different |
| 00:21 | <martinisoft> | ive avoided linux mainly because of the range of distros and packaging systems |
| 00:21 | <martinisoft> | I'm mainly looking for a clean and easy to maintain distro |
| 00:22 | -!- | marvinp [~46bce8d5@webuser.linode.com] has joined #linode |
| 00:22 | <guinea-pig> | apt-get install happiness |
| 00:24 | <booja> | umm, no I had to install the dev tools |
| 00:25 | <booja> | and man |
| 00:25 | <booja> | and syslog |
| 00:25 | <booja> | and a bunch of other stuff |
| 00:25 | <booja> | like I said, slackpkg is greaty |
| 00:25 | <Talman> | Wow. |
| 00:25 | <Talman> | No syslog? |
| 00:25 | <martinisoft> | so theres a lot of stuff to get started I take it |
| 00:25 | <Talman> | Sounds like a stage 2 gentoo. |
| 00:25 | <martinisoft> | any good getting started posts in the forums? |
| 00:26 | <booja> | slackpkg install syslogd |
| 00:26 | <booja> | not too hard to fix at least |
| 00:28 | <martinisoft> | so to get started on slack, what should I do the first thing it comes up? |
| 00:30 | <Bdragon> | enter username |
| 00:30 | <Bdragon> | enter password |
| 00:30 | <Bdragon> | *rimshot* |
| 00:30 | <booja> | get cookies |
| 00:31 | <martinisoft> | Bdragon: way ahead of ya there =p |
| 00:31 | <martinisoft> | booja: check |
| 00:31 | * | Bdragon would opt for "install cvs and bootstrap pkgsrc" |
| 00:31 | <booja> | http://slackpkg.org/ |
| 00:31 | <booja> | install slackpkg |
| 00:32 | <booja> | and point it at a slackware mirror |
| 00:32 | <martinisoft> | its not there by default? |
| 00:33 | <Bdragon> | Heh, slackware was REAL fun back when the system wget was broken due to missing libs... I think I ended up pasting stuff into a terminal and uudecoding it... |
| 00:33 | -!- | ondrej [~ondra@ip4-83-240-41-73.cust.nbox.cz] has quit [Quit: Leaving] |
| 00:34 | <booja> | no, it's not a default package |
| 00:34 | * | JasonF is about to write a nagios check |
| 00:35 | <JasonF> | for io tokens |
| 00:36 | <booja> | io io so off to work I go |
| 00:37 | -!- | SNy [4e9e509e3a@bmx-chemnitz.de] has quit [Read error: Connection reset by peer] |
| 00:37 | <martinisoft> | got it installed |
| 00:38 | <booja> | give it a mirror to point at, in my case I use http://mirrors.usc.edu/pub/linux/distributions/slackware/slackware-12.0/ |
| 00:39 | <booja> | it's in /etc/slackpkg/mirrors |
| 00:40 | <booja> | then do slackpkg update to pull in the package lists |
| 00:40 | <booja> | then just slackpkg search name |
| 00:40 | <booja> | easy peasy |
| 00:40 | <martinisoft> | so why slackpkg over installpkg? |
| 00:41 | <booja> | installpkg only installs packages that you have already downloaded |
| 00:41 | <booja> | slackpkg downloads them, then runs installpkg |
| 00:42 | <martinisoft> | ah nice |
| 00:42 | <martinisoft> | no dep checking though still right? |
| 00:42 | <martinisoft> | i'd need pkg-src for that |
| 00:42 | -!- | atourino [~Antonio@201.218.81.244] has left #linode [Leaving.] |
| 00:44 | <booja> | no not really |
| 00:45 | <booja> | but that's ok, I haven't had too many issues working out what I needed |
| 00:45 | <booja> | oh the other cool thing, you can tell it to install an entire disk set |
| 00:45 | <martinisoft> | interesting |
| 00:46 | <martinisoft> | will it scream at me about dependency issues at least so I can resolve them? |
| 00:48 | <booja> | don't recall |
| 00:48 | <booja> | sorry |
| 00:49 | <martinisoft> | its ok |
| 00:49 | <booja> | really though, if you want dep checking, you want a different distro :P |
| 00:49 | <martinisoft> | well dep checking on those is a double edged sword |
| 00:49 | <@linbot> | New news from forums: IO Token checking nagios plugin in Performance and Tuning <http://www.linode.com/forums/viewtopic.php?t=3151>; |
| 00:50 | <martinisoft> | I'd like one that checks, but doesnt do it for me unless I explicitly tell it so |
| 00:53 | -!- | ^GaveUp^ [gaveup@your.friendly.neighborhood.hellmouth.info] has joined #linode |
| 00:55 | -!- | TheFirst [gaveup@your.friendly.neighborhood.hellmouth.info] has quit [Ping timeout: 480 seconds] |
| 00:55 | -!- | lakin [~lakin@S01060013101832cd.cg.shawcable.net] has joined #linode |
| 00:55 | -!- | lakin [~lakin@S01060013101832cd.cg.shawcable.net] has quit [] |
| 01:05 | <martinisoft> | thanks again for the help |
| 01:05 | -!- | martinisoft [~martiniso@97.102.106.98] has quit [Quit: bbl] |
| 01:15 | -!- | ^GaveUp^ [gaveup@your.friendly.neighborhood.hellmouth.info] has quit [Ping timeout: 480 seconds] |
| 01:27 | <INternat> | so.. |
| 01:27 | <INternat> | network issues? |
| 01:28 | <iggy> | none to dallas or fremont |
| 01:30 | <INternat> | hmms |
| 01:31 | <INternat> | catch.our-lan.com had issues talking to broken.our-lan.com .. i can nenver remember whichi s in which dc |
| 01:32 | <iggy> | I should have said right then |
| 01:33 | <iggy> | I don't know about earlier |
| 01:33 | <iggy> | there was something up in Dallas day before yesterday I think |
| 01:33 | <iggy> | if you are referring to that far back |
| 01:33 | <INternat> | im refering to about 10-15 mins ago |
| 01:33 | <INternat> | unforunatly i dont have a mtr trace |
| 01:34 | <iggy> | I didn't notice anything, but I'm not logged into either server right now |
| 01:35 | <INternat> | i need to get sms notifications set back up so i can get a better response and log in and look |
| 01:35 | <iggy> | actually I am logged in to fremont and it didn't drop the connection |
| 01:35 | <iggy> | or just have your monitoring software run mtr |
| 01:35 | <INternat> | yeah, im usting nagois2 |
| 01:36 | <INternat> | i got hits on pop/smtp/ftp failing |
| 01:37 | -!- | cmarcelo [~cmarcelo@189.70.25.198] has quit [Quit: Leaving] |
| 01:38 | <INternat> | i suppose i could possibly make it run a command when something fails.. * ponders* |
| 01:41 | <encode> | http://bash.org/?845468 |
| 01:42 | <INternat> | heheh yeah ive read that |
| 01:42 | <iggy> | you just had to mention bash.org |
| 01:42 | <iggy> | now I'll be up till 5am hitting random |
| 01:43 | <encode> | heh |
| 01:43 | <encode> | it was an accidental paste anyway - sorry |
| 01:48 | -!- | pfein-away [~pfein@dsl017-025-234.chi1.dsl.speakeasy.net] has quit [Ping timeout: 480 seconds] |
| 01:55 | -!- | marvinp [~46bce8d5@webuser.linode.com] has quit [Remote host closed the connection] |
| 01:56 | -!- | webPragmatist [~webPragma@adsl-99-148-165-44.dsl.rcsntx.sbcglobal.net] has joined #linode |
| 02:23 | -!- | Schroeder [1000@kntpin04-nas-02-s250.cinergycom.net] has quit [Ping timeout: 480 seconds] |
| 02:43 | -!- | row [row@87-194-37-143.bethere.co.uk] has joined #linode |
| 02:50 | -!- | irgeek [~irgeek@cealsa01.centamin.com] has joined #linode |
| 02:58 | -!- | ankur [3bb498db@67.207.141.120] has joined #linode |
| 03:15 | -!- | sdjv100 [~sdjv100@milan.cs.york.ac.uk] has joined #linode |
| 03:15 | -!- | sdjv100 is now known as sveiss |
| 03:33 | -!- | andrew_j_w [~andrew@88-97-29-168.dsl.zen.co.uk] has quit [Quit: Konversation terminated!] |
| 04:01 | -!- | dpn` [~tripped@203.212.17.18] has joined #linode |
| 04:08 | -!- | webPragmatist [~webPragma@adsl-99-148-165-44.dsl.rcsntx.sbcglobal.net] has quit [Quit: Leaving.] |
| 04:23 | -!- | irgeek [~irgeek@cealsa01.centamin.com] has quit [Read error: Connection reset by peer] |
| 04:23 | -!- | irgeek [~irgeek@cealsa01.centamin.com] has joined #linode |
| 04:24 | -!- | sveiss [~sdjv100@milan.cs.york.ac.uk] has quit [Remote host closed the connection] |
| 04:29 | -!- | glenk [~glenk@ip68-4-66-44.oc.oc.cox.net] has quit [Ping timeout: 480 seconds] |
| 04:32 | -!- | mattbnz [~mattb^@home.mattb.net.nz] has quit [Quit: mattbnz] |
| 04:32 | -!- | mattbnz [~mattb^@home.mattb.net.nz] has joined #linode |
| 04:34 | -!- | getsmart [~getsmart@88-149-240-207.dynamic.ngi.it] has joined #linode |
| 04:43 | -!- | INternat [~nf@123-243-184-161.static.tpgi.com.au] has quit [Ping timeout: 483 seconds] |
| 04:55 | -!- | pfein-away [~pfein@dsl017-025-234.chi1.dsl.speakeasy.net] has joined #linode |
| 04:56 | -!- | pfein-away is now known as pfein |
| 04:57 | <rick111> | ;D |
| 04:58 | <pfein> | awake? |
| 04:58 | <pfein> | how do i config postifx on ubuntu? |
| 04:58 | <rick111> | hehe! ask google |
| 04:59 | <irgeek> | pfein: The files are in /etc/postfix |
| 05:10 | <@linbot> | New news from forums: Syslog monitoring in General Discussion <http://www.linode.com/forums/viewtopic.php?t=3152>; |
| 05:11 | -!- | pfein [~pfein@dsl017-025-234.chi1.dsl.speakeasy.net] has quit [Ping timeout: 480 seconds] |
| 05:12 | -!- | metaperl [~metaperl@cpe-75-187-102-204.insight.res.rr.com] has quit [Ping timeout: 480 seconds] |
| 05:12 | -!- | SNy [0101b25449@bmx-chemnitz.de] has joined #linode |
| 05:21 | -!- | getsmart [~getsmart@88-149-240-207.dynamic.ngi.it] has quit [Quit: Ex-Chat] |
| 05:47 | -!- | agentbleubleu [~agentbleu@lns-bzn-38-82-253-92-232.adsl.proxad.net] has joined #linode |
| 05:47 | <agentbleubleu> | morning all |
| 05:48 | <agentbleubleu> | just having my morning oom attack |
| 05:49 | <agentbleubleu> | http://pastebin.linode.com/592 this is the top, http://pastebin.linode.com/593 starting at 5.30 it starts to go oom. |
| 05:51 | <agentbleubleu> | I have removed as many modules from enabled as poss and stopped all unnecessary modules in virtualmin. |
| 05:53 | <encode> | simple solution? add extra ram |
| 05:55 | <agentbleubleu> | yes, i will when its availible next week |
| 05:55 | <agentbleubleu> | but something (cron)? is making it have intermittent fits |
| 05:56 | <irgeek> | It sounds to me like there's a memory leak. My Linode 360 runs all services for half a dozen domains and my swap in/out are zero. |
| 05:56 | <irgeek> | I'd suggest watching memory & swap usage over time to see how it's changing. |
| 05:56 | <agentbleubleu> | i do, it changes dramatically |
| 05:57 | <encode> | agentbleubleu: you're not running any java apps, or a GUI, are you? |
| 05:57 | <agentbleubleu> | with no regard to traffic |
| 05:57 | <agentbleubleu> | no |
| 05:57 | <irgeek> | Do you have graphs of how it's changing? |
| 05:57 | <agentbleubleu> | virtualmin is what i use to control it |
| 05:58 | <agentbleubleu> | in virtualmin the graphs dont change, |
| 05:58 | <agentbleubleu> | from the panel at linode I have some, but they just show cpu and that does not seem to crash at the same time |
| 05:59 | <agentbleubleu> | but cpu has been up at 100 % in last day |
| 05:59 | <agentbleubleu> | in one short spike |
| 06:00 | <agentbleubleu> | in top there are loads of these ksoftirqd/0 migration/3 |
| 06:06 | -!- | Dreamer3 [~Dreamer3@74-142-174-231.dhcp.insightbb.com] has joined #linode |
| 06:10 | -!- | metaperl [~metaperl@cpe-75-187-102-204.insight.res.rr.com] has joined #linode |
| 06:10 | -!- | hpj [~hpj@p54BE93E8.dip0.t-ipconnect.de] has quit [Ping timeout: 480 seconds] |
| 06:11 | -!- | hpj [~hpj@p54BE93E8.dip0.t-ipconnect.de] has joined #linode |
| 06:24 | -!- | metaperl [~metaperl@cpe-75-187-102-204.insight.res.rr.com] has quit [Ping timeout: 480 seconds] |
| 06:24 | <rick111> | anyone ever had this when using mod_security on centos 5? /etc/httpd/modules/mod_security2.so: undefined symbol: ap_get_server_banner |
| 06:25 | -!- | Minty [~workstati@li24-203.members.linode.com] has joined #linode |
| 06:25 | -!- | metaperl [~metaperl@cpe-75-187-102-204.insight.res.rr.com] has joined #linode |
| 06:26 | <Minty> | Hi, is there a graphics driver that wuld work on X on a 360 linode that would perform better than the default? |
| 06:26 | <Minty> | *would |
| 06:27 | -!- | Internat [~nf@123-243-184-161.static.tpgi.com.au] has joined #linode |
| 06:28 | <irgeek> | Minty: UML doesn't have a video card, so there's no graphics driver. |
| 06:28 | <Minty> | I see, thanks anyway |
| 06:31 | <irgeek> | If you're using an x server + vnc to access your Linode, what x server are you using? |
| 06:36 | <Minty> | x.org? The ubuntu package |
| 06:36 | <Minty> | anyway, I'm at school atm and have to go :) |
| 06:36 | -!- | Minty [~workstati@li24-203.members.linode.com] has left #linode [] |
| 06:45 | -!- | mattbnz [~mattb^@home.mattb.net.nz] has quit [Ping timeout: 480 seconds] |
| 06:47 | <rick111> | anyone know what mod, or install I need so 'ap_get_server_banner' is NOT undefined? |
| 06:54 | -!- | mattbnz [~mattb^@home.mattb.net.nz] has joined #linode |
| 06:56 | -!- | metaperl [~metaperl@cpe-75-187-102-204.insight.res.rr.com] has quit [Ping timeout: 480 seconds] |
| 07:19 | <booja> | it's a perl thing, yes? |
| 07:19 | <booja> | http://marc.info/?l=apache-modperl-cvs&m=117761125303579&w=2 |
| 07:19 | -!- | scipio [~scipio@61.6.65.198] has quit [Ping timeout: 480 seconds] |
| 07:22 | <rick111> | think so, let me look at that page, I have come across that page in my searches toda |
| 07:26 | -!- | dpn` [~tripped@203.212.17.18] has quit [Quit: Leaving] |
| 07:27 | <rick111> | fixed it, thanks |
| 07:31 | -!- | TheFirst [gaveup@your.friendly.neighborhood.hellmouth.info] has joined #linode |
| 07:48 | -!- | atourino [~Antonio@201.218.81.244] has joined #linode |
| 07:56 | -!- | hp1 [~hpj@p54BE8AA7.dip0.t-ipconnect.de] has joined #linode |
| 08:01 | -!- | hpj [~hpj@p54BE93E8.dip0.t-ipconnect.de] has quit [Ping timeout: 480 seconds] |
| 08:08 | -!- | TheFirst [gaveup@your.friendly.neighborhood.hellmouth.info] has quit [Ping timeout: 480 seconds] |
| 08:17 | -!- | hp1 [~hpj@p54BE8AA7.dip0.t-ipconnect.de] has quit [Ping timeout: 480 seconds] |
| 08:19 | -!- | linville [~linville@sapphire.tuxdriver.com] has joined #linode |
| 08:22 | -!- | metaperl [~metaperl@cpe-75-187-102-204.insight.res.rr.com] has joined #linode |
| 08:37 | * | caker rematerializes |
| 08:38 | <praetorian> | ok, send in the noobs |
| 08:40 | -!- | ondrej [~ondra@ip4-83-240-41-73.cust.nbox.cz] has joined #linode |
| 08:43 | -!- | Irssi: #linode: Total of 105 nicks [3 ops, 0 halfops, 0 voices, 102 normal] |
| 08:49 | -!- | h00s_ [~h00s@83-131-86-132.adsl.net.t-com.hr] has joined #linode |
| 08:50 | -!- | ankur [3bb498db@67.207.141.120] has quit [Quit: http://www.mibbit.com ajax IRC Client] |
| 08:51 | -!- | h00s [~h00s@83-131-90-200.adsl.net.t-com.hr] has quit [Ping timeout: 480 seconds] |
| 08:53 | -!- | mattbnz [~mattb^@home.mattb.net.nz] has quit [Ping timeout: 480 seconds] |
| 08:54 | <mikegrb> | lolz |
| 08:54 | <mwalling> | lol |
| 08:55 | <mikegrb> | lolz |
| 08:59 | -!- | Bass10 [~fake@c-76-113-194-7.hsd1.mn.comcast.net] has joined #linode |
| 09:02 | -!- | mattbnz [~mattb^@home.mattb.net.nz] has joined #linode |
| 09:02 | -!- | ankur [3bb48463@67.207.141.120] has joined #linode |
| 09:20 | <rick111> | h3l0 |
| 09:22 | <@caker> | oicu812 |
| 09:23 | <mwalling> | roast beef |
| 09:24 | <@caker> | !avail |
| 09:24 | <@linbot> | caker: Linode360 - 41, Linode540 - 2, Linode720 - 0, Linode1080 - 0, Linode1440 - 0 |
| 09:29 | -!- | cmarcelo [~cmarcelo@200.184.118.132] has joined #linode |
| 09:37 | -!- | atourino [~Antonio@201.218.81.244] has left #linode [Leaving.] |
| 09:42 | -!- | getsmart [~getsmart@88-149-240-207.dynamic.ngi.it] has joined #linode |
| 10:01 | -!- | Dreamer3 [~Dreamer3@74-142-174-231.dhcp.insightbb.com] has quit [Quit: Leaving] |
| 10:02 | <mwalling> | has anyone ever set dovecot up to do a mixed local/virtual setup? |
| 10:08 | -!- | Bass10 [~fake@c-76-113-194-7.hsd1.mn.comcast.net] has quit [Ping timeout: 480 seconds] |
| 10:14 | -!- | Bass10 [~fake@c-76-113-194-7.hsd1.mn.comcast.net] has joined #linode |
| 10:17 | -!- | hpj [~hpj@p54BE8AA7.dip0.t-ipconnect.de] has joined #linode |
| 10:25 | -!- | atourino [~Antonio@200.75.192.82] has joined #linode |
| 10:29 | -!- | lakin [~lakin@S01060013101832cd.cg.shawcable.net] has joined #linode |
| 10:29 | <irgeek> | mwalling: Can't the local delivery just be one of the virtual domains? |
| 10:29 | <mwalling> | not sure |
| 10:29 | <mwalling> | better question: can i use both /etc/passwd and a mysql table for auth? |
| 10:33 | <irgeek> | Oh, that. If dovecot is using PAM you can tell PAM to try both. Otherwise, I'm not sure. |
| 10:33 | <iggy> | afaik, yes |
| 10:33 | <mwalling> | ick |
| 10:33 | <mwalling> | pam |
| 10:44 | -!- | lakin [~lakin@S01060013101832cd.cg.shawcable.net] has quit [Ping timeout: 480 seconds] |
| 10:48 | -!- | Dreamer3 [~Dreamer3@74-142-174-231.dhcp.insightbb.com] has joined #linode |
| 10:49 | -!- | pfein [~pfein@dsl017-025-234.chi1.dsl.speakeasy.net] has joined #linode |
| 10:51 | -!- | busycc [~c90c105d@webuser.linode.com] has joined #linode |
| 10:51 | <busycc> | Hi, |
| 10:55 | -!- | lakin [~lakin@S01060013101832cd.cg.shawcable.net] has joined #linode |
| 10:56 | <rick111> | if I'm using crontab to auto load a page which requires authentication, how do I pass the authentication? |
| 10:56 | <rick111> | http://localhost|user:pass or something like that I guess? |
| 10:57 | <@caker> | wget has args to pass that stuff, too |
| 10:57 | <JDLSpeedy> | I think its like http://username:pass@localhost/ |
| 10:57 | <Bdragon> | ^^ |
| 10:57 | <rick111> | thank you! |
| 10:57 | <Bdragon> | http://en.wikipedia.org/wiki/URI_scheme#Generic_syntax |
| 11:00 | -!- | zeph [~zeph@host86-78-dynamic.4-87-r.retail.telecomitalia.it] has joined #linode |
| 11:01 | <irgeek> | Or put it into a file and feed that to wget so the auth info doesn't show up in ps |
| 11:01 | -!- | busycc4 [~busycc@201.12.16.93] has joined #linode |
| 11:02 | <JDLSpeedy> | ahh ya, 'wget `cat file`' |
| 11:02 | <JDLSpeedy> | file would contain http://username:pass@localhost/ |
| 11:02 | <mikegrb> | I keep all of my auth info on a billboard on the way home from work |
| 11:02 | <Bdragon> | I vaugely remember something about putting it in your .netrc as well... |
| 11:02 | <mikegrb> | that way I don't forget anything |
| 11:02 | <JDLSpeedy> | billboard? |
| 11:03 | <mikegrb> | lolz |
| 11:03 | <JDLSpeedy> | LOL |
| 11:03 | <Bdragon> | Might only be used for ftp creds though |
| 11:03 | <busycc4> | Hi, |
| 11:03 | <Bdragon> | hmm |
| 11:03 | <Bdragon> | Nope, looks like wget honors netrc for http as well |
| 11:03 | <busycc4> | any of support? |
| 11:03 | <Bdragon> | and you can use .wgetrc as well as .netrc |
| 11:03 | <busycc4> | bill support... |
| 11:04 | <JDLSpeedy> | busycc4: there around somewhere |
| 11:04 | -!- | pfein is now known as pfein-away |
| 11:04 | -!- | getsmart [~getsmart@88-149-240-207.dynamic.ngi.it] has quit [Ping timeout: 480 seconds] |
| 11:04 | <Bdragon> | mikegrb and caker are active in channel at the moment |
| 11:04 | <Bdragon> | mikegrb, where went your |
| 11:05 | <Bdragon> | +o? |
| 11:05 | -!- | pfein-away is now known as pfein |
| 11:05 | <irgeek> | JDLSpeedy: bash expands stuff in backticks before starting spawning a process |
| 11:05 | <JDLSpeedy> | ooo |
| 11:05 | <Bdragon> | Safest route is probabaly the .netrc / .wgetrc method |
| 11:05 | <mikegrb> | Bdragon: I am incognito |
| 11:05 | <JDLSpeedy> | guess that example don't work, heh |
| 11:06 | <JDLSpeedy> | and chmod it to 600 |
| 11:06 | <Bdragon> | Hmm, that's an interesting idea: |
| 11:06 | <Bdragon> | Quoting the manpage: |
| 11:06 | <Bdragon> | If the passwords are really important, do not |
| 11:06 | <Bdragon> | leave them lying in those files either---edit the files and delete |
| 11:06 | <Bdragon> | them after Wget has started the download. |
| 11:07 | <Bdragon> | also, you can set the WGETRC env var to override where it looks for .wgetrc... |
| 11:08 | -!- | pfein is now known as pfein-away |
| 11:08 | -!- | getsmart [~getsmart@88-149-230-191.dynamic.ngi.it] has joined #linode |
| 11:08 | <Bdragon> | so you can write the creds to an arbitrary file and then WGETRC=~/foo wget http://..... |
| 11:11 | -!- | pfein-away is now known as pfein |
| 11:12 | <agentbleubleu> | ref my oom prob, in my syslog i see it started with http://pastebin.linode.com/593 Mar 13 05:30:48 li6-87 kernel: oom-killer: gfp_mask=0x201d2, order=0 . I have found this : http://bugs.centos.org/view.php?id=1391 which suggests a patch. This is way out of my depth, could anyone pass comment? |
| 11:13 | -!- | busycc4 [~busycc@201.12.16.93] has quit [Quit: busycc4] |
| 11:15 | <mikegrb> | agentbleubleu: that patch says it is to fix a bug in 64bit systems which your Linode is not |
| 11:15 | <agentbleubleu> | thanks mike |
| 11:16 | <Bdragon> | How about not letting the system run out of memory? ;) |
| 11:16 | <@caker> | Free swap = 0kB |
| 11:16 | <@caker> | agentbleubleu: something is eating all your VM |
| 11:16 | <@caker> | agentbleubleu: it's not a kernel problem |
| 11:16 | <scott> | hungry hungry hippos |
| 11:17 | <@caker> | da da da da da daaaa! |
| 11:17 | <agentbleubleu> | Bdragon: would i do that by lowering the php memory limit? its now at 32Mg? |
| 11:18 | <Bdragon> | No, you need to figure out why you're eating through all your ram and swap and tune the system to not do that. |
| 11:18 | <Bdragon> | run top? ;) |
| 11:18 | <agentbleubleu> | top is constantly in my view |
| 11:18 | <Bdragon> | Sorting by memory %? |
| 11:18 | <agentbleubleu> | sec |
| 11:20 | <agentbleubleu> | sorry for being an idiot but how to sort by memory |
| 11:20 | <@caker> | cap M |
| 11:20 | <Bdragon> | Press > a few times |
| 11:21 | <agentbleubleu> | thanks |
| 11:22 | <agentbleubleu> | mysql at top allways |
| 11:22 | <irgeek> | agentbleubleu: Do you have procinfo loaded? If not, grab the package and see what your swap in & swap out are. |
| 11:22 | <irgeek> | On a well behaved Linode, they shouldn't change often. |
| 11:22 | <Bdragon> | What's RES and %MEM of mysqld? |
| 11:22 | <agentbleubleu> | 1259 mysql 15 0 140m 18m 2420 S 7 3.3 6:32.78 mysqld |
| 11:23 | <agentbleubleu> | PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND |
| 11:23 | <agentbleubleu> | 1259 mysql 15 0 140m 17m 2420 S 0 3.3 6:34.15 mysqld |
| 11:23 | <@caker> | agentbleubleu: cat /proc/swaps, please? |
| 11:24 | <agentbleubleu> | Filename Type Size Used Priority |
| 11:24 | <agentbleubleu> | '/dev/xvdb partition 262136 55512 -1 |
| 11:24 | <agentbleubleu> | ' |
| 11:25 | <@caker> | ok .. 55M in swap |
| 11:25 | <@caker> | obviously, it would be easier to identify the offending process once your swap usage is close to full |
| 11:26 | <agentbleubleu> | maybe this might help, here it was when at 100% swap: http://pastebin.linode.com/592 |
| 11:27 | <Bdragon> | High load average |
| 11:27 | <agentbleubleu> | yes |
| 11:28 | <Bdragon> | Offending process(es) memory wise isn't on the list though.. |
| 11:28 | <agentbleubleu> | http://pastebin.linode.com/593 |
| 11:28 | <agentbleubleu> | this was the oom at the time |
| 11:28 | <Bdragon> | Got any cron jobs that run at that time? |
| 11:28 | <agentbleubleu> | i looked no |
| 11:28 | <agentbleubleu> | 1 hour beofre |
| 11:28 | <agentbleubleu> | or after |
| 11:28 | <Bdragon> | had it been running for an hour? ;) |
| 11:28 | -!- | Schroeder [1000@kntpin04-nas-02-s127.cinergycom.net] has joined #linode |
| 11:28 | <agentbleubleu> | no |
| 11:29 | <agentbleubleu> | like 24 hour |
| 11:29 | <agentbleubleu> | s |
| 11:29 | <Bdragon> | How about apache access log? |
| 11:29 | <agentbleubleu> | sec |
| 11:31 | <mwalling> | agentbleubleu still OOMing? |
| 11:31 | <agentbleubleu> | yep |
| 11:32 | <Bdragon> | If it spikes again try to run SHOW FULL PROCESSLIST in mysql as well... |
| 11:32 | <agentbleubleu> | ok |
| 11:33 | <agentbleubleu> | i will run that now to see how it works |
| 11:33 | <Bdragon> | Oh, put a colon after it ;) |
| 11:33 | <Bdragon> | err, semicolon |
| 11:33 | <agentbleubleu> | access logs look weird, |
| 11:33 | <agentbleubleu> | yes ok |
| 11:34 | <Bdragon> | httpd looks weird immediately before the oom? |
| 11:34 | <mwalling> | crazy idea, might do more harm then good, but how about emailing an offsite email account the output of 'ps aux' every n minutes? |
| 11:35 | <mwalling> | then you could flip through them and see whats gobbling it up? |
| 11:35 | <Bdragon> | (How about every 1 minute if the load is over 5? ;) |
| 11:35 | <mwalling> | that would work too |
| 11:36 | <mwalling> | mikegrb: is there a dhcpd answering on 192.168.128.0/17? |
| 11:36 | <agentbleubleu> | http://pastebin.linode.com/594 |
| 11:36 | <agentbleubleu> | this is SHOW FULL PROCESSLIST; now |
| 11:37 | <mikegrb> | mwalling: not yet |
| 11:37 | <mwalling> | k |
| 11:37 | <Bdragon> | Using persistent connections? |
| 11:37 | <mwalling> | j/w |
| 11:37 | <Bdragon> | Hmm |
| 11:37 | <agentbleubleu> | no i dont think so |
| 11:37 | <Bdragon> | Those are short sleeps |
| 11:38 | <Bdragon> | Run it a few more times |
| 11:38 | <agentbleubleu> | but i am not using mysql_close |
| 11:39 | <Bdragon> | Also... |
| 11:39 | <agentbleubleu> | http://pastebin.linode.com/595 |
| 11:39 | <Bdragon> | USE mplaylist; SHOW KEYS FROM playlist; |
| 11:40 | <agentbleubleu> | mysql> USE mplaylist; SHOW KEYS FROM playlist; |
| 11:40 | <agentbleubleu> | ERROR 1049 (42000): Unknown database 'mplaylist' |
| 11:40 | <agentbleubleu> | ERROR 1046 (3D000): No database selected |
| 11:40 | <agentbleubleu> | mysql> USE mplaylist; |
| 11:40 | <agentbleubleu> | ERROR 1049 (42000): Unknown database 'mplaylist' |
| 11:40 | <agentbleubleu> | mysql> SHOW KEYS FROM playlist; |
| 11:40 | <agentbleubleu> | ERROR 1046 (3D000): No database selected |
| 11:40 | <agentbleubleu> | mysql> |
| 11:40 | <agentbleubleu> | ? |
| 11:40 | <@caker> | ... |
| 11:40 | <Bdragon> | err, myplaylist.. Just wondering why a simple update was showing up in the processlist ;) |
| 11:41 | <Bdragon> | Might have just been a timing thing on your part ;) |
| 11:41 | -!- | glenk [~glenk@ip68-4-66-44.oc.oc.cox.net] has joined #linode |
| 11:42 | -!- | standard [~597aff20@webuser.linode.com] has joined #linode |
| 11:42 | <Bdragon> | also, STATUS; |
| 11:42 | <Bdragon> | The last row.. |
| 11:43 | <agentbleubleu> | Threads: 4 Questions: 145575 Slow queries: 141 Opens: 564 Flush tables: 1 Open tables: 64 Queries per second avg: 17.489 |
| 11:43 | <Bdragon> | That's a lot of slow queries for only 145k questions |
| 11:43 | <Bdragon> | Are you using table indexes appropriately? |
| 11:43 | -!- | standard0 [~standard@89.122.255.32] has joined #linode |
| 11:44 | -!- | standard0 [~standard@89.122.255.32] has quit [] |
| 11:44 | <Bdragon> | BTW, I meant "USE myplaylist; SHOW KEYS FROM playlist;" -- I typoed the db name when I asked before |
| 11:44 | <agentbleubleu> | im not 100% sure on anything. I am using a class called freetag which is a tagging system that is thridpartie |
| 11:45 | -!- | chesty [~chesty@mail.chesterton.id.au] has quit [Remote host closed the connection] |
| 11:45 | <Bdragon> | You should definately have at least an index on id, and probabaly a pk |
| 11:46 | <agentbleubleu> | but one of the scripts is really heavy, it takes about 2 minutes to run normally |
| 11:46 | <Bdragon> | What does it do? |
| 11:46 | <mwalling> | um... |
| 11:46 | <mwalling> | that sounds wasteful |
| 11:46 | <Bdragon> | If it takes two minutes, you most likely are missing an index or two |
| 11:46 | <agentbleubleu> | it does a lot, it tries to find a similar band to the last one played. |
| 11:47 | <Bdragon> | if we're talking about two minutes waiting on the db to crunch |
| 11:47 | <agentbleubleu> | its a long script |
| 11:47 | <mwalling> | agentbleubleu: means nothing |
| 11:47 | <mwalling> | its wasting resources... looking throught the queries it makes, you could prolly add an index and get much better results |
| 11:48 | -!- | standard [~597aff20@webuser.linode.com] has quit [Remote host closed the connection] |
| 11:48 | -!- | glen1 [~glenk@ip68-4-66-44.oc.oc.cox.net] has joined #linode |
| 11:48 | <Bdragon> | You can get many thousands of times performance increase in most cases between a table missing useful indexes and having proper ones. |
| 11:49 | -!- | glenk [~glenk@ip68-4-66-44.oc.oc.cox.net] has quit [Ping timeout: 480 seconds] |
| 11:49 | <Bdragon> | (i.e. dropping from a 100 second query to 0.2 seconds, etc) |
| 11:49 | <agentbleubleu> | aha |
| 11:50 | <agentbleubleu> | ok i describe what it does then you might understand more: |
| 11:50 | <Bdragon> | Additionally, you might want to turn on slow query logging |
| 11:50 | <agentbleubleu> | is that in my.cnf |
| 11:50 | <agentbleubleu> | i think i did that yesterday |
| 11:50 | <Bdragon> | Would be interested to see the slow query log |
| 11:51 | <agentbleubleu> | ok sec |
| 11:51 | <mikegrb> | Would be interested to see the Bdragon |
| 11:51 | <mikegrb> | oops |
| 11:51 | <Bdragon> | heh |
| 11:51 | <mikegrb> | did I say that out loud? |
| 11:51 | <Bdragon> | I'm right here! |
| 11:51 | -!- | hfb [~hfb@pool-71-118-254-245.lsanca.dsl-w.verizon.net] has joined #linode |
| 11:52 | <Bdragon> | I think it's actually a command line argument.. http://dev.mysql.com/doc/refman/5.0/en/slow-query-log.html |
| 11:52 | <mikegrb> | anybody around with multiple linodes in dallas other then mwalling? |
| 11:52 | <agentbleubleu> | http://pastebin.linode.com/596 |
| 11:52 | <mwalling> | mikegrb: or i'll talk to them if they want |
| 11:52 | -!- | glenk [~glenk@ip68-4-66-44.oc.oc.cox.net] has joined #linode |
| 11:53 | <Bdragon> | Heh |
| 11:53 | <Bdragon> | THAT looks like a missing index. |
| 11:53 | <mikegrb> | mwalling: hmm? |
| 11:53 | <Bdragon> | DESCRIBE SELECT * FROM playlist WHERE approved = '1'; |
| 11:53 | <mikegrb> | oh |
| 11:53 | <mikegrb> | heh |
| 11:53 | -!- | glenk [~glenk@ip68-4-66-44.oc.oc.cox.net] has quit [Read error: Connection reset by peer] |
| 11:53 | <Bdragon> | after doing a USE myplaylist; |
| 11:53 | -!- | glenk [~glenk@ip68-4-66-44.oc.oc.cox.net] has joined #linode |
| 11:54 | <agentbleubleu> | im not sure what you mean by a missing index |
| 11:54 | <Bdragon> | I mean you don't have appropriate indexes on the tables. It's an SQL thing. |
| 11:54 | -!- | glenk [~glenk@ip68-4-66-44.oc.oc.cox.net] has quit [Read error: Connection reset by peer] |
| 11:55 | <Bdragon> | USE myplaylist; DESCRIBE SELECT * FROM playlist WHERE approved = '1'; |
| 11:55 | -!- | glenk [~glenk@ip68-4-66-44.oc.oc.cox.net] has joined #linode |
| 11:55 | <Bdragon> | (DESCRIBE will print the execution plan of the query) |
| 11:55 | <agentbleubleu> | ok |
| 11:56 | <agentbleubleu> | http://pastebin.linode.com/597 |
| 11:56 | -!- | glen1 [~glenk@ip68-4-66-44.oc.oc.cox.net] has quit [Ping timeout: 480 seconds] |
| 11:56 | <Bdragon> | Yep, that's a table scan all right |
| 11:57 | <Bdragon> | ALTER TABLE playlist ADD INDEX approved (approved); |
| 11:57 | <Bdragon> | DESCRIBE SELECT * FROM playlist WHERE approved = '1'; |
| 11:58 | <agentbleubleu> | its taking time |
| 11:58 | <agentbleubleu> | mysql> ALTER TABLE playlist ADD INDEX approved (approved); |
| 11:58 | <agentbleubleu> | Query OK, 12265 rows affected (56.97 sec) |
| 11:58 | <agentbleubleu> | Records: 12265 Duplicates: 0 Warnings: 0 |
| 11:59 | <agentbleubleu> | http://pastebin.linode.com/598 |
| 11:59 | <Bdragon> | Much better |
| 12:00 | <mikegrb> | now try SELECT * FROM playlist WHERE approved = '1'; |
| 12:00 | <Bdragon> | don't do that |
| 12:00 | <mikegrb> | ignore the output but look at the query execution time at the end |
| 12:00 | <mikegrb> | Bdragon: :< |
| 12:00 | <Bdragon> | heh |
| 12:00 | <mikegrb> | Bdragon: that's the best part! |
| 12:00 | * | mikegrb thwaps Bdragon |
| 12:00 | <agentbleubleu> | easy there |
| 12:00 | <Bdragon> | Well, if you have a relatively fast connection you can do that |
| 12:00 | <agentbleubleu> | i have |
| 12:00 | <Bdragon> | but it's gonna spam 5011 rows |
| 12:00 | <agentbleubleu> | ? |
| 12:01 | <agentbleubleu> | pls dont fuck up me db |
| 12:01 | <Bdragon> | so be expecting that if you do run the actual select |
| 12:01 | <Bdragon> | I mean it's gonna be a really long resultset |
| 12:01 | <agentbleubleu> | aha |
| 12:01 | <Bdragon> | I'm on satellite so I kinda tend to flinch when someone suggests running a query that returns a resultset like that :P |
| 12:01 | <agentbleubleu> | can i just ask what you just did already |
| 12:02 | <mwalling> | Bdragon: heh |
| 12:02 | <mwalling> | you'll get fapped! |
| 12:02 | <Bdragon> | I had you add an appropriate index for doing the "SELECT * FROM playlist WHERE approved = '1'" query |
| 12:02 | <Bdragon> | I was fapped yesterday. |
| 12:02 | <mwalling> | ouch |
| 12:02 | <Bdragon> | Just came out of it ~1h ago |
| 12:02 | <Bdragon> | stupid steam updates |
| 12:02 | <mwalling> | get back from MA and forget? |
| 12:03 | <Bdragon> | There's still room for improvement on the table |
| 12:03 | <Bdragon> | Actually, much room. |
| 12:03 | <Bdragon> | DESCRIBE playlist; |
| 12:03 | <Bdragon> | I'm very interested in seeing how the id field is set up... |
| 12:04 | <agentbleubleu> | http://pastebin.linode.com/599 |
| 12:04 | <Bdragon> | ok, that's keyed at least. Good. |
| 12:04 | <agentbleubleu> | the main 2 are the long blobs |
| 12:04 | <agentbleubleu> | they hold large arrays |
| 12:05 | <Bdragon> | # |
| 12:05 | <Bdragon> | # Query_time: 16 Lock_time: 0 Rows_sent: 1 Rows_examined: 12186 |
| 12:05 | <Bdragon> | # |
| 12:05 | <Bdragon> | SELECT * FROM playlist WHERE id LIKE '7061'; |
| 12:05 | <Bdragon> | # |
| 12:05 | <agentbleubleu> | one is of picture urls and the other mp3 urls |
| 12:05 | <Bdragon> | I think this one is just a case of using LIKE instead of = accidentally and the query optimizer getting confused |
| 12:05 | <Bdragon> | DESCRIBE SELECT * FROM playlist WHERE id LIKE '7061'; |
| 12:05 | <Bdragon> | DESCRIBE SELECT * FROM playlist WHERE id = '7061'; |
| 12:05 | * | mwalling notes that Bdragon posesses great mysql-fu |
| 12:06 | <mikegrb> | mwalling: nah he had good google skills |
| 12:06 | * | mikegrb urns |
| 12:06 | <agentbleubleu> | http://pastebin.linode.com/600 |
| 12:06 | <Bdragon> | Nah |
| 12:06 | <Bdragon> | I do a lot of optimization on mysql stuff |
| 12:06 | <mwalling> | mikegrb is an urn? |
| 12:06 | <mwalling> | heh |
| 12:06 | * | mikegrb nods |
| 12:07 | <agentbleubleu> | one of the arrays normally holds 500 urls. |
| 12:07 | <Bdragon> | yeah, the first one is a table scan because it wasn't able to use the primary key because you used LIKE, which IIRC doesn't work on numeric fields.. |
| 12:07 | <Bdragon> | (I mean, the index doesn't work when doing a LIKE on numeric fields) |
| 12:07 | <agentbleubleu> | i did nt know that |
| 12:08 | <Bdragon> | And the second query should have PRIMARY for "key" on the second query |
| 12:08 | <Bdragon> | (DESCRIBE SELECT * FROM playlist WHERE id = '7061';) |
| 12:09 | <Bdragon> | If this is correct, it would be worthwhile to figure out where in the code you're using LIKE against an int field ;) |
| 12:09 | <agentbleubleu> | http://pastebin.linode.com/601 |
| 12:09 | <Bdragon> | Yep |
| 12:09 | <Bdragon> | OK, now for fun: |
| 12:10 | <Bdragon> | Run the two queries and compare the execution time ;) |
| 12:10 | <Bdragon> | (i.e. run them without the DESCRIBE in front) |
| 12:10 | <agentbleubleu> | ok |
| 12:11 | <agentbleubleu> | first one had a fit |
| 12:11 | <agentbleubleu> | 1 row in set (16.13 sec) |
| 12:12 | <Bdragon> | If second one is over 0.2 seconds I will frankly be suprised. |
| 12:12 | <agentbleubleu> | seems stuck |
| 12:12 | <agentbleubleu> | frozen |
| 12:12 | <agentbleubleu> | i check top |
| 12:13 | <agentbleubleu> | 1 row in set (1 min 21.50 sec) |
| 12:13 | <agentbleubleu> | load average: 8.98, 5.17, 2.80 |
| 12:13 | <Bdragon> | You in the middle of spiking? |
| 12:13 | <Bdragon> | SHOW FULL PROCESSLIST; |
| 12:13 | <@caker> | also, ps auxhfwww |
| 12:15 | <agentbleubleu> | yea spiking |
| 12:15 | <agentbleubleu> | http://pastebin.linode.com/602 |
| 12:15 | <agentbleubleu> | not getting response at mo |
| 12:15 | <@caker> | is top sorting by memory still? |
| 12:16 | <agentbleubleu> | i just make sure i think sp |
| 12:16 | <agentbleubleu> | here we go |
| 12:16 | <agentbleubleu> | http://pastebin.linode.com/603 |
| 12:17 | <Bdragon> | lock contention.. |
| 12:17 | <agentbleubleu> | top is still by memory |
| 12:17 | <Bdragon> | Hmm, sorting things.. |
| 12:18 | <Bdragon> | Looks like it's piling up behind a lock |
| 12:18 | <@caker> | I blame thrashing. |
| 12:18 | <Bdragon> | Yeah, probabaly |
| 12:19 | <@caker> | agentbleubleu: step 1. Figure out what's eating memory. Step 2. fix it |
| 12:19 | <agentbleubleu> | what is thrashing |
| 12:19 | <Bdragon> | The linode |
| 12:19 | <@caker> | agentbleubleu: swap thrashing |
| 12:19 | <anderiv> | step 3: profit! oh wait.... |
| 12:19 | <agentbleubleu> | i changed swappiness to 20 |
| 12:19 | <Bdragon> | I still blame keys |
| 12:19 | <Bdragon> | # |
| 12:19 | <Bdragon> | # |
| 12:19 | <Bdragon> | | 3809 | myplaylist | localhost | myplaylist | Query | 181 | Sorting result | select * from playlist where approved = '1' ORDER BY id DESC LIMIT 0, 5 | |
| 12:19 | <@caker> | agentbleubleu: watch top, sorted my memory usage, Swap used is > 200, capture the full output of top |
| 12:20 | <@caker> | *if swap used is |
| 12:20 | <agentbleubleu> | 20%? |
| 12:20 | <@caker> | who what? |
| 12:20 | <@caker> | 200 MB |
| 12:20 | <mwalling> | turkey. |
| 12:21 | <agentbleubleu> | i changed swappenis to 20% i think |
| 12:21 | <agentbleubleu> | sec i get my notes |
| 12:21 | <Bdragon> | Can you shutdown apache? |
| 12:21 | <Bdragon> | At least then it would quit piling more queries on.. |
| 12:21 | <@caker> | agentbleubleu: swappiness is irrelevant in this scenario |
| 12:21 | <agentbleubleu> | i can do what ever you want |
| 12:21 | <@caker> | agentbleubleu: if top says that you're using over 200 MB of swap space, capture the full output |
| 12:21 | <agentbleubleu> | ok |
| 12:22 | <Bdragon> | If you can, stop apache temporarily |
| 12:22 | <@caker> | agentbleubleu: and pastebin it. |
| 12:22 | <@caker> | too many cooks |
| 12:22 | <Bdragon> | I think your users are hitting refresh because the site is unresponsive, thereby contributing to the meltdown.. |
| 12:22 | <@caker> | All we need to do is figure out who's going overboard and limit it. It's easy |
| 12:23 | <agentbleubleu> | caker please describe how to capture the full output |
| 12:23 | <Bdragon> | (I rather think ironing out the indexes will solve most of the problem) |
| 12:23 | <@caker> | agentbleubleu: copy paste? |
| 12:23 | <@caker> | Bdragon: if mysql doens't have limits on it, there's still the potential to blow up |
| 12:23 | <agentbleubleu> | sec |
| 12:24 | <@caker> | Bdragon: indexes reduce the amount of disk scanning required to complete a query |
| 12:24 | <agentbleubleu> | http://pastebin.linode.com/604 |
| 12:24 | <agentbleubleu> | this is top by memory |
| 12:24 | <Bdragon> | Yeah, but at 17 qps on the server, anything that reduces full table scans gives a huge boost |
| 12:24 | <@caker> | Bdragon: yes, they can reduce memory usage (and increase it), but if there are no limits to mysql the problem will still exist |
| 12:24 | <agentbleubleu> | is this what you meant |
| 12:24 | <Bdragon> | It's most likely apache waiting on mysqld... |
| 12:25 | <@caker> | agentbleubleu: I want that when the machine is spiraling down |
| 12:25 | <@caker> | agentbleubleu: or better yet, |
| 12:25 | -!- | lakin_ [~lakin@S01060013101832cd.cg.shawcable.net] has joined #linode |
| 12:25 | <@caker> | agentbleubleu: or better yet, "ps auxhfwwww" when the machine is spiraling down |
| 12:25 | <Bdragon> | pid 7763 looks rather interesting.. That's quite a difference in memory usage.. |
| 12:25 | <agentbleubleu> | i have it when it went oom |
| 12:26 | <agentbleubleu> | but not organised by memory |
| 12:26 | <@caker> | then it's useless |
| 12:26 | <agentbleubleu> | ok |
| 12:26 | <@caker> | "ps auxhfwwww" <-- when machine is thrashing is all we need |
| 12:26 | <agentbleubleu> | ok |
| 12:26 | <agentbleubleu> | asap |
| 12:27 | <agentbleubleu> | Bdragon: pid 7763 looks rather interesting.. That's quite a difference in memory ? |
| 12:27 | <Bdragon> | Could you try "SELECT * FROM playlist WHERE id = '7061';" again now that it's not thrashing... |
| 12:27 | <agentbleubleu> | sure |
| 12:27 | <Bdragon> | I mean, it's using a lot more ram than the others |
| 12:27 | <Bdragon> | It's probabaly serving a page that is being real mean |
| 12:27 | <agentbleubleu> | 1 row in set (2.37 sec) |
| 12:27 | <Bdragon> | A bit better |
| 12:28 | <Bdragon> | If the db were healthy it should be nearly instantaneous |
| 12:28 | <agentbleubleu> | so what have i fucked up using like instead of = |
| 12:28 | <agentbleubleu> | and not enough indexes (which i dont understand) |
| 12:28 | -!- | Schroeder [1000@kntpin04-nas-02-s127.cinergycom.net] has quit [Ping timeout: 480 seconds] |
| 12:28 | <Bdragon> | Well, that's a big part of it |
| 12:29 | <Bdragon> | Let's fix the indexes and then find that LIKE... |
| 12:29 | <agentbleubleu> | ok |
| 12:29 | <agentbleubleu> | the like i can find |
| 12:29 | <agentbleubleu> | i know it all like the back of my hand |
| 12:29 | <Bdragon> | OK |
| 12:29 | <Bdragon> | http://pastebin.linode.com/603 is very useful regarding seeing what's not using indexes properly and why stuff is fighting, btw |
| 12:30 | <agentbleubleu> | aha |
| 12:30 | <Bdragon> | Found the LIKE already? |
| 12:31 | <agentbleubleu> | these queries without anythink in them look bloody odd |
| 12:31 | <agentbleubleu> | give me 2 minutes i do that now |
| 12:31 | <agentbleubleu> | SELECT * FROM playlist WHERE id LIKE '' |
| 12:31 | <agentbleubleu> | this shouldnt occur as far as I have designed t |
| 12:31 | <Bdragon> | Doesn't matter really, an = will let it at least use the index (and reduce the load on the db drastically) |
| 12:31 | <agentbleubleu> | ok |
| 12:31 | <agentbleubleu> | i do the likes now |
| 12:31 | <agentbleubleu> | pls give me a few minutes |
| 12:32 | <Bdragon> | k |
| 12:32 | <mwalling> | Bdragon: yeah, lay off :P |
| 12:32 | <Bdragon> | I'm gonna make some orange juice, I seem to be coming down with something |
| 12:32 | <mikegrb> | it workde! |
| 12:32 | -!- | lakin [~lakin@S01060013101832cd.cg.shawcable.net] has quit [Ping timeout: 480 seconds] |
| 12:37 | <agentbleubleu> | found 1 |
| 12:37 | <agentbleubleu> | looking for more |
| 12:39 | <agentbleubleu> | can onnly find one |
|
