[00:14] <SpamapS> shade34321: apache's config is extremely complex. Its hard to tell what is wrong without understanding your whole config.
[00:15] <shade34321> SpamapS: ok, thanks. I don't even know the entire config, took it over and haven't had a chance to look at it much, if you have any questions about it I will try my best to answer them
[00:17] <lifeless> SpamapS: and even then .... ..
[00:21] <shade34321> does apache have a irc channel?
[00:22] <SpamapS> shade34321: You probably want to make sure that there are no 'Options' lines with 'Indexes' in the listanywhere
[00:24] <shade34321> hmm...question...Options Indexes and Options -Indexes are not the same thing are they?
[00:25] <philpem> shade34321, first one enables, second one disables
[00:25] <philpem> most specific takes priority
[00:25] <shade34321> <--idiot
[00:25] <shade34321> thanks
[00:26] <philpem> no, the word you're looking for is "newbie", or possibly "greenhorn".
[00:26] <philpem> the only stupid question is the one you didn't ask.
[00:26] <SpamapS> shade34321: Options -Indexes should turn it off
[00:26] <shade34321> I've been looking at this to long...lol...I kept reading about -Indexes and then seeing Indexes and not seeing the difference
[00:26] <shade34321> that it did
[00:26] <shade34321> now to see if everythign else works right
[00:26] <shade34321> o.O
[00:26] <SpamapS> shade34321: the real question is why did it get turned on.
[00:26] <philpem> on the specificity rule, if you have Options Indexes set for /var/www, then clear it for /var/www/images, images will be unindexed but everything else will be indexed
[00:26] <shade34321> thanks for the help!
[00:27]  * SpamapS suggests version control for configs.. *soon*.. ;)
[00:27] <philpem> SpamapS, some web developers turn it on to be lazy... I have it set on my development servers, and for only one directory (temp) on the release server
[00:27] <shade34321> <--took over last may and I haven't touched any config stuff and the people who did do this graduated and will not answer questions as to why certain things are the way they are
[00:27] <philpem> well.. release server == my website
[00:28] <philpem> I already have version control for configs -- I use Mercurial to do it :)
[00:28] <shade34321> i've actually had a lot of questions for them since my knowledge is fragmented to get told to google it
[00:28] <shade34321> :)
[00:28] <philpem> hg addrem / hg commit before I change anything in /etc, then hg commit after I change it. something goes wrong, I hg up -C to the rev before I changed it.
[00:28] <philpem> I can roll everything back to when I last installed or upgraded if I need to.
[00:29] <philpem> Or trundle through the change history and see what I changed, when, and why.
[00:30] <philpem> "Don't explain what you did, you can get that from the change history and deltas. Instead, explain what the problem was and why you fixed it that way."
[00:31] <philpem> "Years after you leave, your colleagues will praise you for your foresight, and thank the Gods of Computing that you were working on their codebase. Instead of, you know, wishing you a nasty end involving sharks with frickin' laser beams."
[00:32] <Kyle__> philpem: Sure you're not talking about my predecessor at this job? For the last part.
[00:33] <shade34321> or mine?
[00:33] <Kyle__> philpem: But you forgot, sharks with laser beams, and syphallis.
[00:33] <shade34321> lol
[00:35] <Kyle__> I'm setting up a server with multiple-heads, to put up interesting graphs of what's going on on the network (manager distractors).  Does ubuntu-server's X-config work just like the desktop config?
[00:35] <philpem> My goal at work: to have people remember me as "the one guy in the SE department who bothered to document his code, and write decent commit comments"
[00:35] <philpem> Kyle__, should do.
[00:35] <philpem> install ubuntu-desktop for GNOME3, kubuntu-desktop for KDE, xubuntu-desktop for XFCE.
[00:35] <philpem> that'll pretty much turn an Ubuntu Server box into a desktop
[00:36] <philpem> or just pick the app you want and install it... you can run X without a window manager but it isn't fun :)
[00:36] <Kyle__> philpem: Lofty goal.  I was teased heavily for 6 months about commiting ever peice of documentation to a wiki.  Then cursed by other departments when it was documented that we did _EVERYTHING_ right on our side, thank you very much.  Then thanked heartily when I left, for leaving the new sysadmin with viable, up-to-date documentation :)
[00:36] <philpem> if you want a manager distractor and don't care for functionality, you probably want GNOME3 (he says, only half joking, and with a grin like the Cheshire Cat's)
[00:38] <Kyle__> philpem: Heh.  Gnome3 would eat up all the CPU.  I'm going to generate some little graphs via a lightweight ruby script and toss them up as X backgrounds, rotating through the monitors periodically.
[00:38] <philpem> Kyle__, You haven't seen my code... 1:1 code:comment ratio, and anything with a public API has Doxygen comments.
[00:39] <philpem> you can throw Doxygen at it, and get fully searchable and *useful* documentation.
[00:39] <Kyle__> Wowzers.
[00:40] <philpem> every so often I read through the docs and make sure I haven't made any silly mistakes, so there are commits and CRs along the lines of "Fix stupid documentation mistake: open_device_fast does NOT return an int, it returns an E_CT_ERROR_CODE!"
[00:41] <philpem> My manager used to ask why I was "filling the CR database up" with those. Until one of the other developers asked him to pass along a message: "Can you ask Phil if he's got time to document ES_CP_VIDEO_ENGINE? I know it's not his code, but his documentation rocks!"
[00:41] <Kyle__> Sweeet.
[00:50] <SpamapS> Kyle__: re the manager distractor.. if I were in your situation, I'd do all the graphics w/ HTML and just run a browser fullscreen.
[00:50] <SpamapS> Kyle__: backgrounds and X programming seem like overkill for a simple rotating image. Also, ratpoison should suffice for your window management needs. :)
[00:55] <Kyle__> SpamapS: Oh, not real X programming, just tossing up a background with a CLI program.  Backgrounds in X are held in graphics memory, and take up zero CPU.  Can't say that about firefox or webkit, even when they're idle.
[00:55] <Kyle__> SpamapS: Ratpoison?  Humm.  Don't you need something to put that on, so the manager eats it?
[00:55] <SpamapS> Kyle__: how crappy is this box you're putting this on that you're worried about most likely about 20MB of memory?
[00:56] <SpamapS> Kyle__: and if you do it w/ HTML you can let the managers run it on their desks, and not even come over to look at the distractors ;)
[00:57] <Kyle__> SpamapS: Not crappy at all, but it's going to serve as the head of an openstack cluster.  If my experience with eucalpytus is a judge, this system will occasionaly swap under load, up until I get ~24GB of ram into it.
[00:58] <SpamapS> Kyle__: do not compare openstack to eucalyptus so quickly. ;) openstack is the anti-euca when it comes to scaling
[00:58] <Kyle__> SpamapS: God I hope so.  That's why we switched when moving from our proof-of-concept hardware to real hardware.
[00:58]  * Kyle__ loves academia: you couldn't pull a change like that in business so quickly.
[00:59] <SpamapS> err
[00:59] <SpamapS> thats so not true of successful businesses.
[00:59] <shade34321> Kyle__: what is this cluster going to be used for if you don't mind me asking?
[01:00] <SpamapS> Kyle__: successful businesses know when to stop shovelling money into a hole just as fast as academics.
[01:00] <Kyle__> shade34321: For students, it will run web-apps, or whatever server-code they're writing for class.  1-1 or 1-many student to vm ratio.  Also will be used for sysadmin classes.  MOre importantly, I'll spin down the student VMs at night, and be running hadoop and nexus jobs on it :)
[01:01] <shade34321> nice
[01:01] <Kyle__> SpamapS: Yes, but usually you'd have to re-do the proof of concent with the new software, before getting the new hardware.  At least that's what I've seen.
[01:01] <shade34321> at my job I'm in charge of 3 HPC clusters, two running slurm, one running pbs
[01:02] <SpamapS> Kyle__: depends on how burdened with big and clunky red tape the business is.
[01:02] <Kyle__> shade34321: pbs?
[01:02] <shade34321> one sec and ill give a link
[01:03] <Kyle__> SpamapS: and how flush the manager running the project is feeling, coupled with their feelings towards the employee/contractor.
[01:04] <Kyle__> These Dell C5000s are neat.  Like having 12 blades in 2 U.  Sadly without the fancy network interconnect, but a fraction of the price.
[01:06] <Kyle__> Sorry, 3 u.
[01:06] <shade34321> sorry....to many links
[01:06] <Kyle__> Heh.
[01:06] <shade34321> stands for Portable Batch System and it's used in unison with torque and maui
[01:07] <shade34321> it's for academic research, AE to be exact
[01:07] <shade34321> currently the boss wants to get a new cluster since our main one is dying:(
[01:07] <Kyle__> Neat.
[01:08] <Kyle__> Shame mosix isn't still going strong.
[01:08] <shade34321> mosix?
[01:08] <Kyle__> How big of a cluster, if I can pry?
[01:08] <Kyle__> Rigth now I'm feeling lucky to have these 96 cores, but I know that's small potatoes.
[01:09] <shade34321> one of them is about to be decommissioned, it's really old and we've lost 60% of the nodes, one is 256 nodes with 2 dual cores procs(it's an IBM machine from early 2000's) and the other is a 30 node cluster with dual quad core xeons
[01:09] <SpamapS> mosix was a mess
[01:09] <SpamapS> neat idea
[01:09] <SpamapS> but too naive
[01:09] <shade34321> lol
[01:10] <Kyle__> SpamapS: Yea, but if they hadn't tanked, it may have been georgous by now.
[01:10] <shade34321> our big cluster, with 1024 cores is currently at 214 active nodes at 856 cores:(
[01:11] <shade34321> his current path is he wants to stick as many cores per node as he can
[01:11] <Kyle__> shade34321: Yea.  Small university, smaller budget.  But if this pans out the budgets of next year and years after may get funneled right into my racks!
[01:11] <shade34321> so 4 socket boards:D
[01:11] <shade34321> understandable
[01:11] <SpamapS> Kyle__: I'm not sure I agree.. the idea ignored things like map/reduce entirely and focused just on trying to magically make multi-processing scale out.
[01:11] <Kyle__> shade34321: Wow does that sound fun.
[01:12] <shade34321> the new opterons though seem a flop, our origianl idea so it'd be 64 cores per board
[01:12] <shade34321> *seem to be a flop
[01:12] <SpamapS> Kyle__: hadoop and its map/reduce cousins work because they acknowledge that big data requires specific strategies to break it up.
[01:12] <Kyle__> SpamapS: Which for some jobs is the easiest way of comprehending them.  If you can get it "good-enough" it's worth it to trade some computational efficiency for programmer efficiency.
[01:12]  * Kyle__ nods
[01:12] <shade34321> and we're currently talking over the new intel ones with 10 cores
[01:13] <Kyle__> shade34321: The i series xenons?  Why did I think they were at 12?
[01:13] <shade34321> b/c they should be?
[01:13] <SpamapS> I think all the mosix-friendly problems can be handled simply by writing generic workers and communicating jobs to them via message queues.
[01:13]  * SpamapS got really excited about mosix when he first saw it tho
[01:14] <shade34321> since we're talking about clusters...i have a question for you guys
[01:14] <Kyle__> SpamapS: Especially compared to the PVM code I did for an undergrad project, it seemed like mana from hevan.
[01:15] <shade34321> our clusteres are running RHEL or CentOS(same difference) and currently we don't seem to have any temp monitoring software/hardware in place
[01:15] <SpamapS> Yeah, definitely nicer to just fork and run than have to get into PVM muck
[01:15] <shade34321> what do you guys recomend for it?
[01:16] <Kyle__> shade34321: You tried the traditional i2c sensor package?  I forget what it's called in centos, been awhile since I tried it.  It never recognized the hardware in the Dells at my last gig, but once in awhile it would on nicer hardware.
[01:17] <shade34321> well our dells our r410's with an IBM x3550 mgt node, home built cluster built before I arrived
[01:17] <shade34321> and the IBM are all IBM x3550 nodes or some variant, don't reemember exact numbers
[01:17] <shade34321> wanted to use the iDRAC stuff from Dell but the head node stands in my way:/
[01:17]  * SpamapS shuts down for the day
[01:17] <Kyle__> shade34321: See if the dells came with a DRAC or iDRAC.  Dell Remote Access Console.  They usually share the first NIC, and have a web-page that you can power-off or on the box from, and give sensor readouts.
[01:18] <Kyle__> Heh.
[01:18] <SpamapS> Kyle__: good luck. There are a lot of openstack experts in here and in #openstack ... don't hesitate to ask. :)
[01:18] <Kyle__> SpamapS: Thanks!
[01:18] <shade34321> Kyle__: I couldn't find iDRAC installed but again I don't want to manually check all 30 nodes
[01:18] <Kyle__> shade34321: Get into the network, behind the head-node.  See if the DRAC/iDRACs in your dells supporrt SNMP.  If they do, you should be able to pull the sensor data using snmpget from the head-node.
[01:19] <Kyle__> is there a DHCP server?
[01:19] <shade34321> I want to say it's the head node
[01:19] <Kyle__> Hey, some people hate DHCP (though I don't know why)
[01:20] <Kyle__> Check for a bunch of DHCP requests that aren't your nodes.  Could be DRACs.  On some dells you have to enable it from the BIOS first though.
[01:20] <shade34321> I'll take a look into it tmr afternoon when I stop by
[01:21] <shade34321> and see what I can find, though last time I looked I couldn't find a drac option on start up...though I did not check the bios
[01:21] <Kyle__> Good luck.  I should head out myself
[01:21]  * Kyle__ waves
[01:21] <shade34321> thanks
[01:21] <shade34321> night
[04:26] <twb> Does dnsmasq have an upstream VCS?   I can't find it.
[06:32] <xInterlopeR777x> !ubuntu
[06:33] <twb> re dnsmasq -- never mind, I isolated the fault on my own.
[06:50]  * rmk is away: no reason supplied
[06:50]  * rmk is no longer away
[08:12] <Vivek> Hi
[08:12] <Vivek> I am facing issues with Enlisting a system at install time with the Ubuntu Orchestra Server.
[08:12] <Vivek> The step it fails is select and install software.
[08:16] <shang> Vivek: are you doing that from the server CD?
[09:22] <Vivek> Yes I am doing it from the server cd.
[09:25] <Vivek> The ubuntu client does not get authenticated to the ubuntu orchestra server.
[09:46] <eagles0513875> hey guys i am a bit lost im trying to create a new install on a remote server via kvm and i want to setup lvm i have a root partition which isnt part of the lvm and 960gb of free space which i want to setup as a logical volume
[09:47] <eagles0513875> what are the steps to do this as i am a bit lost
[09:47] <koolhead17> hi all
[10:09] <atari2600a> hey
[10:09] <atari2600a> what's the default location for KVM images?
[10:10] <atari2600a> I'm setting up a server on my desktop & I want to give the VM its own partition & the more vanilla I do it now the easier it'll be when I end up wiping my / partition whenever
[10:11] <koolhead17> !virt-manager
[10:12] <koolhead17> atari2600a: https://help.ubuntu.com/community/KVM/VirtManager  see if this is what you need
[10:12] <atari2600a> please don't be cryptic
[10:12] <koolhead17> !virtmanager
[10:12] <atari2600a> it's a simple question & if I can get this done BEFORE I create the image it'll make everyone's life easier
[10:15] <jamespage> morning
[10:16] <Daviey> hey jamespage !!
[10:16] <jamespage> gudday Daviey!
[10:17] <atari2600a> I'm...gonna go
[10:40] <lynxman> morning o/
[10:41] <lynxman> jamespage: morning sir ;)
[10:41] <koolhead17> hey Daviey jamespage  :)
[10:41] <koolhead17> elhoo lynxman
[10:44] <lynxman> Daviey: morning! happy new year ;)
[10:46] <Daviey> hey lynxman, trust you had a good one?
[10:47] <lynxman> Daviey: internet disconnection for 2 weeks, it was very good indeed, how about you?
[10:51] <Daviey> lynxman: wow, how did you cope?
[10:52] <lynxman> Daviey: I did indeed :)
[10:53] <lynxman> Daviey: didn't know I could but it definitely was refreshing :)
[11:05] <Daviey> lynxman: sounds like an ordeal
[11:06] <lynxman> Daviey: :)
[12:46] <zul> morning
[13:24] <lynxman> zul: morningtons
[13:39] <zul> lynxman: not exact internet disconnection though :)
[13:39] <Deathvalley122> is it a bad sign of a bad drive when you get I/O error when trying to set up lvm?
[14:21] <Daviey> SpamapS: Upgradibg oneiric -> precise, mysql-server has asked me for the password 3 times, which i keep leaving blank
[14:22] <Daviey> "If this field is left blank, the password will not be changed"
[15:09] <lynxman> Daviey: Trying to come back slowly, do you remember the nomenclature we were talking about for a package including the git last checkout hash into its name?
[15:09] <lynxman> Daviey: I think it was package-N.N+git~checkouthash right?
[15:11] <Daviey> lynxman: date 20120105-shorthash is better IMO.
[15:12] <lynxman> Daviey: ah that's the one, thanks :)
[15:12] <Daviey> date will always increment in value, hash will not.
[15:12] <Daviey> perhaps s/-/.
[15:13] <lynxman> Daviey: indeed, that was the issue
[15:13] <lynxman> Daviey: building new mcollective-plugins, so the version will be 0.0.0~git20120105-9b90c2b-0ubuntu1
[15:13] <Daviey> lynxman: so one exampe, mythtv (2:0.24.0+fixes.20111207.40f3bae-0ubuntu1)
[15:13] <lynxman> Daviey: would that be okay? :)
[15:13] <lynxman> Daviey: ah ok, dot
[15:14] <Daviey> yeah, but maybe use a . instead of -
[15:14] <Daviey> winner
[15:14] <lynxman> Daviey: :)
[15:14] <stgraber> hallyn: around?
[15:15] <hallyn> stgraber: sneaking off soon for breakfast.  what's up?
[15:16] <stgraber> hallyn: was wondering if you had an idea why we don't get a console on /dev/console with a clean precise container
[15:16] <stgraber> hallyn: getty is spawned but nothing shows up
[15:16] <stgraber> hallyn: also lxc-console seems to behave a bit differently (like if it was resetting the console or something like that)
[15:16] <hallyn> is that new?
[15:16] <hallyn> someone mentioned something similar on lxc-users m-l i think
[15:17] <Daviey> smoser: are you around?
[15:17] <hallyn> stgraber: in short no i've not looked into it
[15:17] <stgraber> hallyn: I only started noticing it a few weeks ago. I don't really use precise containers that much
[15:17] <hallyn> oneiric containers on precise don't do that?
[15:18] <stgraber> hallyn: well, at least lucid containers on precise don't
[15:18] <hallyn> it's possible i messed something up with the console.conf...
[15:20] <hallyn> (creating some new containers to play with)
[15:20] <stgraber> hallyn: found the problem I think, well, at least what caused it. Downgrading util-linux to Oneiric's version fixes it
[15:21] <hallyn> yuck
[15:24]  * stgraber starts diffing the two source packages
[15:38] <yakster> anyone know how to fix "/var/run/dbus/system_bus_socket: No such file or directory" when trying to install mysql?
[15:42] <zul> Daviey: hes on holiday this week
[15:48] <Daviey> zul: who approved THAT?!
[15:49] <zul> gah?
[15:50] <RoAkSoAx> lol
[16:36] <Koheleth> bunt support with webmin ok?
[16:36] <Koheleth> or vice versa come to that
[16:37] <Koheleth> dumping Plesk
[16:38] <Koheleth> worst cp gui there is
[16:42] <EvilResistance> !webmin | Koheleth
[16:48] <pmatulis> Koheleth: no, do not use webmin
[17:19] <SpamapS> Daviey: hrm.. not sure why mysql would bug you repeatedly for a password on upgrade.. it shouldn't even *ask* on upgrade.
[17:19] <SpamapS> Daviey: possible we need to port the mysql-5.1 values forward to 5.5.. hrm
[17:19]  * SpamapS is hrming a lot
[17:23] <SpamapS> no.. we use the generic mysql-server/root_password
[17:23] <SpamapS> so debconf should be silent
[17:50] <adam_g> RoAkSoAx: ping
[17:50] <Daviey> Can everyone make sure blueprints are up to date please? :)
[17:51] <Daviey> rbasak / jamespage: ^^ before you go for the day please? :)
[17:51] <Daviey> (sorry for the late reminder)
[17:52] <RoAkSoAx> adam_g: pong
[17:52] <Daviey> Spamaps,  rbasak and hazmat: http://status.ubuntu.com/ubuntu-precise/group/topic-precise-servercloud-service-orchestration.html
[17:52] <Daviey> ^^ can you update that please :)
[17:53] <SpamapS> Daviey: I'm working on the juju mir right now.. will mark them all as INPROGRESS
[17:54] <RoAkSoAx> Daviey: shgould we start preparing orchestra/cobbler for MIR too or shouild I just wait post sprint?
[17:55] <RoAkSoAx> err rally
[17:55] <adam_g> RoAkSoAx: were you ever able to resolve that grub issue you had yesterday with orchestra? the same thing is blocking precise installs oon the openstack cluster
[17:55] <RoAkSoAx> adam_g: nope, still having the issue as of this morning. I'm waiting for this afternoon to see if it gets solved
[17:57] <SpamapS> RoAkSoAx: MIR's are mostly just gathering of information for the MIR team to digest.. so I'd say go for it, unless you think we won't be MIR'ing cobbler. :)
[17:57] <adam_g> RoAkSoAx: is there a bug for it somewhere?
[17:58] <rbasak> Daviey, nothing to update. I think I've just about cleared most other outstanding stuff, was hoping to make a start with the pandaboard tomorrow.
[18:00] <RoAkSoAx> SpamapS: indeed I guess I'll start then ;)
[18:04] <RoAkSoAx> adam_g: hold on let me retry
[18:05] <smoser> Daviey, i'm here now
[18:06] <adam_g> RoAkSoAx: doing the same
[18:24] <adam_g> RoAkSoAx: same thing
[18:25] <RoAkSoAx> adam_g: I can't seem to find a bug with the issue, so i guess we'll have to file one
[18:26] <adam_g> http://paste.ubuntu.com/794026/
[18:26] <adam_g> RoAkSoAx: syslog ^
[18:27] <RoAkSoAx> adam_g: yeah same error
[18:34] <RoAkSoAx> adam_g: have you tried with i386?
[18:36] <adam_g> RoAkSoAx: i have not
[18:36] <adam_g> RoAkSoAx: filing a bug against grub-installer now
[18:42] <hallyn> grr - the latest precise server mini iso fails on grub install
[18:42] <hallyn> oh heh
[18:43] <adam_g> :)
[18:43] <hallyn> man, lag is insane.  takes liek 30 seconds for chars to show up
[18:44] <adam_g> hallyn: are you doing anything special in the preseed (orchestra / juju-wise) or just a straight up precise install?
[18:46] <hallyn> adam_g: i've had this both with and without preseed
[18:46] <adam_g> hallyn: ok, i guess thats "good"
[18:46] <hallyn> first tried 'vm-new' from the vm-tools, and then just by hand with kvm on cmdline
[18:47] <hallyn> on the bright side, you can just reboot from cd, rescue, and install grub2 by hand :)  so all the more i wonder why it fails in installer
[18:48] <adam_g> hallyn: Bug #912431 if you want to weigh in / confirm
[18:48] <adam_g> RoAkSoAx: ^
[18:48] <hallyn> thanks, will do
[18:54] <hallyn> stgraber: you know, I wonder if the lxc /dev/console issue is related to all my recent server installs popping up on an empty console
[18:55] <hallyn> oh, no, i bet that's bc it still wants to jump the X window even though there is no x
[18:57] <hallyn> yeah, it's the vt.handoff=7 in /proc/cmdline.  getting rid of that fixes that.
[18:59] <hallyn> oddly, that bit isn't in /etc/default/grub.
[19:03] <yakster> hey all, have an issue installing mysql…. says that it cannot connect to dbus
[19:04] <yakster> start: Unable to connect to system bus: Failed to connect to socket /var/run/dbus/system_bus_socket: No such file or directory
[19:04] <SpamapS> yakster: installing?
[19:04] <yakster> ok yeah now its not installing … sorry got ahead of myself
[19:04] <SpamapS> yakster: that should only happen when you try to run 'start mysql' as a non-root user
[19:05] <yakster> can't install mysql-server cause mysql-server-5.1 is depends….
[19:05] <SpamapS> yakster: please explain what you tried to do, what you expected to happen, and what actually happened.
[19:05] <yakster> ok….
[19:07] <yakster> installed mysql, and it failed, told me that
[19:07] <yakster> Package mysql-server-5.1 is not configured yet.
[19:07] <yakster> dpkg: error processing mysql-server (--configure):
[19:07] <yakster>  dependency problems - leaving unconfigured
[19:07] <SpamapS> ok, try 'sudo dpkg --configure -a'
[19:07] <yakster> ok
[19:07] <yakster> Setting up mysql-server-5.1 (5.1.54-1ubuntu4) ...
[19:08] <SpamapS> don't paste it all here
[19:08] <yakster> pastebin is down
[19:08] <yakster> sry
[19:08] <SpamapS> http://paste.ubuntu.com .. :)
[19:09] <yakster> http://paste.ubuntu.com/794071/
[19:09] <smoser> yakster, also, fyi, 'pastebin' command is your friend.
[19:09] <smoser> er... pastebinit.
[19:09] <smoser> utlemming, https://code.launchpad.net/~smoser/ubuntu-on-ec2/ec2-publishing-scripts.detach-vol-cleanups/+merge/87668
[19:10]  * utlemming looks
[19:14] <yakster> what is the pastebinit syntax?
[19:16] <yakster> nm i got it
[19:22] <yakster> SpamapS: anything?
[19:22] <SpamapS> yakster: sorry I got distracted :-P
[19:23] <yakster> np
[19:23] <yakster> says it can't continue because of a previous failure
[19:23] <SpamapS> yakster: ok, look in /var/log/mysql/* .. there may be some clues there
[19:23] <SpamapS> yakster: most likely problem is an error in /etc/my.cnf
[19:23] <SpamapS> err
[19:24] <SpamapS> /etc/mysql/my.cnf
[19:24] <yakster> I "rm -R /etc/mysql" when trying to reinstall from scratch
[19:25] <SpamapS> yakster: that would be the problem then
[19:25] <SpamapS> yakster: dpkg considers that intentional, and will leave the files there
[19:25] <SpamapS> err, leave the files "deleted"
[19:25] <yakster> so what….
[19:25] <SpamapS> yakster: try using 'dpkg -i /var/cache/apt/archives/mysql-server-5.1*.deb --force-confmiss
[19:26] <SpamapS> err, without the extra ' ;)
[19:26] <RoAkSoAx> adam_g: how did you manage to pastebin the syslog from the failed installation?
[19:26] <SpamapS> yakster: next time use 'apt-get purge mysql-server-5.1' if you want to get rid of the config files.
[19:26] <SpamapS> yakster: though that will also remove the databses.
[19:26] <yakster> didnt know that…. but now i do
[19:27] <yakster> ok still mysql-server-5.1, errors were encounterd
[19:28] <SpamapS> yakster: its one of the more confusing things about dpkg... but its intentional since sometimes you do want to rm a config file and not have it come back.
[19:28] <SpamapS> yakster: is /etc/mysql back?
[19:28] <SpamapS> yakster: you may need to do mysql-common as well
[19:28] <yakster> not yes
[19:28] <yakster> ok mysql-common
[19:28] <yakster> ok
[19:29] <yakster> yes, mysql is back
[19:29] <adam_g> RoAkSoAx: there is a menu option to dump debug logs, and the installer serves them to you via http
[19:30] <RoAkSoAx> adam_g: where's that menu option?
[19:30] <adam_g> RoAkSoAx: somewhere at the bottom, near 'execute a shell'
[19:30] <RoAkSoAx> adam_g: ok cool ;)
[19:32] <smoser> utlemming, please review that above.
[19:32] <yakster> , with all that you told me heres what I finally did to get it all to work…. apt-get  purge mysql*
[19:33] <yakster> apt-get install mysql
[19:34] <yakster> and now it is working
[19:34] <utlemming> smoser: merging now
[19:34] <yakster> thank you
[19:37] <Combatjuan> I'm looking at iotop and finding that the Total DISK WRITE is often an order of magnitude or more than the actual sum of disk writes per process.  What does that mean?
[19:39] <Combatjuan> The only thing I can figure is that it means that it's writing to swap, however iostat indicates that swap is not being touched.
[19:46] <SpamapS> Combatjuan: journalling is also an issue
[19:48] <Combatjuan> SpamapS - Interesting.  I'm seeing 20M/s in iotop (that's the total, the combined children come to about 1M/s).  That seems like an awful lot of journalling.  Could it possibly even be anywhere near that high?
[19:49] <SpamapS> Combatjuan: are you using any kernel based services like kernel nfs server maybe?
[19:50] <SpamapS> Combatjuan: also, dumb question.. RAID?
[19:50] <Combatjuan> Not sure about kernel-based services.  Yes, it is RAID.  Not sure why that's a dumb question though.
[19:50] <Combatjuan> (Hardware RAID)
[19:50] <SpamapS> Combatjuan: another possibility.. if your processes are writing *tiny* things.. your disks are writing *blocks* .. so if writes are 400 bytes each, but you have 4k blocks.. that would explain it
[19:51] <SpamapS> especially if things are highly random and not sequential
[19:51] <Combatjuan> SpamapS - If the per-process settings are in fact per byte and the total is a sum of blocks, that would definitely seem like a possibility.
[19:52] <Combatjuan> s/settings/numbers
[19:54] <SpamapS> Combatjuan: I don't know for sure how iotop works, but its at least a theory. :)
[19:55] <SpamapS> Combatjuan: I'd suggest stracing your processes and looking at how big the write's are.. and also looking at how often they fsync (or if the files are opened O_SYNC).
[19:55] <SpamapS> Combatjuan: strace -e trace=fsync,write -p xxxx
[19:56] <Combatjuan> Excellent idea.  Thanks SpamapS.
[19:59] <Combatjuan> It's a postgres server.  It looks like most ever write is exactly one block (which I think is pretty normal).  Sometimes a few blocks at a time.  Each line is an fsync though?
[19:59] <SpamapS> yeah with postgres I'd expect it to be fairly efficient with the writes
[20:00] <SpamapS> Combatjuan: perhaps iotop is misleading you about the total writes of all processes
[20:00] <Daviey> RoAkSoAx: wait for the Rally.
[20:00] <SpamapS> Combatjuan: note.. that the cheapest solution for all write problems (even cheaper than diagnosing like you're doing now) is to go buy a FusionIO card and stick your data on that. ;)
[20:01] <Combatjuan> That's my best guess.  There are times when it claims that it wrote 50M/s.  It seems to use bytes though it doesn't explitely say so everywhere.  I don't think the raid (simple mirror) can even do that.
[20:02] <Kyle__> How do you put a desktop environment on ubuntu-server, and have it only start when you run startx?  I dont' want it running all the time
[20:02] <Combatjuan> SpamapS: Yeah, I do love solving things with hardware even though that's hard to do from half a continent away.  But this is the first time that the server has ever seemed to be disk-bound instead of something else.  So I'm not ready to do that yet.
[20:02] <RoAkSoAx> Daviey: OK
[20:02] <Kyle__> upstart seems so odd to me.
[20:02] <SpamapS> Kyle__: you're not alone. :)
[20:03] <SpamapS> Kyle__: its quite elegant once you get used to it though.
[20:36] <thyrant> hi guys
[20:36] <semiosis> hi all, i have a quesiton about contributing ubuntu-specific improvements to a debian package
[20:37] <semiosis> specifically, how would I go about doing that?
[20:38] <semiosis> the package, glusterfs-server, is included in ubuntu from debian, but there's an ubuntu-specific bug resulting from our use of upstart/mountall.
[20:38] <thyrant> I am thinking about updating my ubuntu server 9.10 .. how much trouble might I be in if I do so?
[20:38] <semiosis> i have an upstart job which solves the problem, which can be added to the packaging
[20:39] <semiosis> see https://bugs.launchpad.net/ubuntu/+source/glusterfs/+bug/876648
[20:43] <Kyle__> thyrant: From 9.10 to 11.10?  They changed what they call the DHCP and DNS servers, even though they're the stock ISC ones in both versions.  I can't recall a whole lotta differences.
[20:45] <thyrant> stock ISC ?
[20:47] <Kyle__> thyrant: internet systems consortium.  They're DHCP server is the standard, and that's what debian ubuntu and most every other distro has always come with
[20:48] <Pici> thyrant: You'll need to go through all the intermediate releases if you upgrade from 9.10 to 11.10.  May I suggest just upgrading to 10.04 and then to 12.04 when it is released (you can update from  LTS to LTS)
[20:48] <Kyle__> thyrant: They also make BIND
[20:48] <thyrant> the server works fine though.. I only wanted to update to get a supported version so I can install bittorrent. Cant figure out how to do it in 9.10
[20:48] <Pici> thyrant: Then just upgrade to 10.04
[20:51] <thyrant> thanks I'll do that
[20:52] <RoAkSoAx> adam_g: do you have any orchestra server with dnsmasq?
[20:53] <RoAkSoAx> in the lab
[21:00] <RoAkSoAx> adam_g: nevermind
[21:09] <cjz> can anyone help with a postfix issue, im trying to get amazon ses working through postfix and id like to up the debugging to see what pipe is actually doing
[21:10] <cjz> ive done -vvvv on the pipe daemon entry but its not showing how its calling the actual argv command
[21:11] <Patrickdk> !verbose
[21:11] <Patrickdk> oh heh, wrong channel
[21:11] <Patrickdk> increasing verbose in postfix is hardly ever needed by anyone
[21:14] <semiosis> cjz: you know amazon recently announced SMTP support for SES?  so your apps can connect directly to SES via SMTP and not have to use a postfix gateway anymore
[21:14] <cjz> my app is nagios
[21:14] <cjz> would that work?
[21:15] <semiosis> i'm not sure, i'm still using the postfix gateway :D
[21:19] <semiosis> cjz: here's my config for example: http://pastie.org/3133795
[21:20] <semiosis> cjz: i found the amazon-provided perl scripts to be completely worthless, so i wrote this nice & clean python script using boto (requires boto 2.0+) to do the ses delivery
[21:21] <semiosis> cjz: however, thinking about this again, since SES now supports SMTP itself, you can probably skip the whole pipe-to-a-script method and just have postfix deliver to another SMTP server which is SES
[21:21] <newbie|2> hi all
[21:22] <newbie|2> i've a question about ldap configuration with proftpd
[21:22] <newbie|2> when i try to connect on my ftp, i've this message : "invalid dn syntaxt" in proftpd.log
[21:23] <newbie|2> syntax*
[21:23] <semiosis> cjz: oops just realized that should be master.cf, not main.cf, in the pastie link i just sent
[21:24] <newbie|2> my ldap server is a windows serveur 2003 active directory
[21:25] <jwac> hello...anyone active?
[21:35] <jwac> yo anyone got a sec?  I have a quick question about nullmailer
[21:37] <SpamapS> jamespage: around?
[21:37] <SpamapS> !ask
[21:37] <SpamapS> jwac: ^^
[21:37] <jwac> k
[21:37] <Kyle__> How long does patience take?
[21:37] <Kyle__> !patience
[21:37] <SpamapS> Kyle__: I'll tell you tomorrow ;)
[21:37] <Kyle__> !impatience
[21:38] <Kyle__> Hu.  It really should have that one....
[21:38] <robbiew> !yell
[21:39] <jwac> I use nullmailer.  Whenever I send an email from the CLI it just sits in my queue, not autosending.  I'm thinking it may be the permissions on the queue folder because I mistakingly deleted it and had to recreate it.  I want to start by asking, can anyone find out what the default permissions on the nullmailer queue folder are for me please?
[21:48] <thyrant> hey guys, me with ubuntu server 9.10 again.. My question is now: what is the best procedure to backup the whole system, repartition the drives and restore the system to exactly the same state as before?
[21:49] <Patrickdk> I thought 9.10 went unsupported a long time ago
[22:03] <Kyle__> Is anyone here airly familiar with dell's DRAC cards?
[22:30] <cjz> woot
[22:30] <cjz> semiosis:  that was it, got it working thank you very much
[22:39] <semiosis> cjz: glad to hear it, you're welcome