[00:09] <RoAkSoAx> SpamapS: yes http://paste.ubuntu.com/583586/
[00:12] <RoAkSoAx> SpamapS: found the error
[00:12]  * RoAkSoAx really needs some eye rest
[00:17] <SpamapS> 1923  bind(12, {sa_family=AF_FILE, path="/var/run/mysql1.sock"}, 110) = -1 EACCES (Permission denied)
[00:17] <SpamapS> that?
[00:17] <SpamapS> Oi.. super late.. have to run
[00:31] <knoxy> hi all.. How can I do to install Ubuntu Server in my HP Proliant DL120 G6 using the Smart Array B110i SATA? I'm trying, but the ubuntu can't detect my logical drives...
[01:07] <iggi> Can anyone help me with an weird apache issue? I can see it listening on port 80, but when I try to view the web page I get The server is taking too long to respond.
[01:07] <twb> iggi: you're testing with "curl -v http://127.0.0.1/" ?
[01:12] <RoyK> omg... someone in .jp tweeted http://transport.nilu.no/products/fukushima
[01:13]  * RoyK is in severe need for sleep
[01:55] <iggi> twb: No, I am trying from a remote machine. It worked fine, I restart the host and then it gives me this
[01:55] <iggi> twb: I tried stop-start, restart. Checked configs, etc.
[02:11] <smoser> kirkland, ubuntu uec image faile dto build with:
[02:11] <smoser> VMBuilder.exception.VMBuilderException: Process (['chroot', '/tmp/vmbuilderwWI9QQ/root',
[02:11] <smoser> 'dpkg-reconfigure', '-fnoninteractive', '-pcritical', 'console-setup']) returned 2.
[02:11] <smoser> stdout: , stderr: md5sum: /usr/share/console-setup/vtrgb: No such file or directory
[02:11] <smoser> update-alternatives: error: alternative path /etc/console-setup/vtrgb doesn't exist.
[02:24] <kirkland> smoser: hmm
[02:24] <kirkland> smoser: okay ... I wonder why that file isn't found
[02:25] <kirkland> smoser: can i see smore logs?
[02:27] <kirkland> smoser: i think this should be easy to solve with some more log
[02:31] <kirkland> smoser: nevermind, i think i see it
[02:38] <kirkland> smoser: uploading a fix
[02:41] <kirkland> smoser: thanks; fix uploaded
[02:46] <smoser> kirkland, well, might be fixed might not be
[02:46] <smoser> i just realized that 0.11 vmbuilder copies /etc/console from the host to the target
[02:46] <smoser> which i think is disgusting
[02:47] <smoser> but works without issue from lucid -> karmic, lucid, maverick, natty (until today)
[02:47] <smoser> or at least without issue that i know of
[02:47] <smoser> oh, and add hardy to that list
[02:47] <kirkland> smoser: hmmf
[02:47] <kirkland> smoser: yeah, that's pretty bad
[03:08] <lifeless> SpamapS: hey
[03:08] <lifeless> SpamapS: can has lucandra?
[03:08] <lifeless> :>
[03:08]  * lifeless is all want want want
[03:10] <SpamapS> lifeless: !WINNING
[03:11] <SpamapS> lifeless: can you point me to the install HOWTO?
[03:14] <lifeless> of course
[03:14] <lifeless> https://github.com/tjake/Lucandra
[03:14] <lifeless> (its really solandra now, but he hasn't renamed it)
[03:14] <lifeless> From the Solandra base directory:
[03:14] <lifeless> mkdir /tmp/cassandra-data
[03:14] <lifeless> ant
[03:14] <lifeless> cd solandra-app; ./start-solandra.sh
[03:15] <SpamapS> heh
[03:15]  * SpamapS starts looking for jars
[03:15] <lifeless> AIUI we'll need to run two clusters
[03:15] <SpamapS> lifeless: I've been having trouble w/ Cassandra 0.7.4 .. it doesn't like one of the lib versions in maverick/natty
[03:15] <lifeless> one random partitioner
[03:15] <lifeless> and one solandra w/ordered partitioner
[03:15] <SpamapS> ordered because it will be time-archived right?
[03:16] <lifeless> so it may make sense to just have different packages with different data dirs etc - so that devs can have a 1-node cluster of both by apt-get install
[03:16] <SpamapS> lifeless: I think *all* of its dependencies except cassandra are already in the archive.. +++
[03:17] <SpamapS> of course.. they've been very explicit about versions.. because java people don't know how to do forward/backward compatibility.. so hopefully there's an extensive test suite. :-P
[03:17] <lifeless> SpamapS: I'll drop matt a mail asking to confirm this, concrete data is a little hard to get
[03:19] <SpamapS> ahh no such luck.. needs thrift too
[03:25] <SpamapS> lifeless: hopefully somewhere between puppet, lxc, and ensemble, we can have a dev story that involves a more realistic environment. :)
[03:27] <lifeless> SpamapS: mmmm
[03:27] <twb> Didn't you know?  "Enterprise" means you ship a distro worth of deps in /opt, so you know you get the tested versions of same!
[03:27] <lifeless> SpamapS: -very- unconvinced
[03:28] <twb> What's ensemble?
[03:28] <lifeless> SpamapS: love ensemble, but test suites requiring -that stack- - not keen
[03:29] <SpamapS> lifeless: what stack?
[03:30] <SpamapS> twb: service management tool..  https://launchpad.net/ensemble
[03:30] <lifeless> SpamapS: lxc etc
[03:30] <SpamapS> lifeless: lxc is integrated fully into natty and can run lucid as a guest w/ minimal tweaking..
[03:30] <lifeless> SpamapS: 'make check' spinning up containers, debugging, etc - not going to be pretty
[03:31] <SpamapS> lifeless: I dunno.. its being written with just this sort of story in mind
[03:31] <lifeless> SpamapS: I'm sure it /can/ be done; the question is why would you want to.
[03:31] <lifeless> SpamapS: so not interesting. Fantastic deployment story.
[03:31] <lifeless> dev cycle story, I just don't get it.
[03:31] <SpamapS> lifeless: to stop the cascade of failures that happen when dev assumptions cause QA/production fail
[03:31] <twb> That's an Canonical-driven project?
[03:31] <SpamapS> twb: its an experimental thing righ tnow
[03:32] <twb> OK, then I stop caring :-)
[03:32] <SpamapS> twb: been under dev for a while.. just now functional
[03:32] <lifeless> SpamapS: thats a pipeline scenario, not a unit test scenario
[03:32] <twb> I can't think of much Canonical has funded that I actually like
[03:32] <lifeless> twb: not even Ubuntu ?
[03:32] <SpamapS> lifeless: I tend to think the closer you can make a dev's environment to reality, the less translation problems you will have.
[03:32] <twb> I hate ubuntu
[03:33] <lifeless> twb: ...
[03:33] <SpamapS> fine line between love and hate
[03:33] <twb> I use it because 1) management/customers won't go for Debian; and 2) at least it isn't RHEL/CentOS/SLES.
[03:33] <lifeless> SpamapS: if you do a pareto analysis on LP severe problems, its lack of automation + high latency reporting, not being 'like production'
[03:33] <SpamapS> twb: to be fair, I'm fairly certain that you don't like *anything*
[03:33] <twb> Oh definitely
[03:34] <lifeless> SpamapS: remember, we can't use production data (data protection act), and to detect problems you *need* 300GB of DB.
[03:34] <lifeless> SpamapS: I want test suites that can run a single test in < 3-4 *seconds*
[03:34] <SpamapS> lifeless: agreed its hard to detect log(n) scaling problems when n is 2. ;)
[03:34] <lifeless> SpamapS: I just don't see a big stack, no matter how good, participating in that story.
[03:35] <SpamapS> lifeless: the stack would already be running the pieces that don't need updating...
[03:35] <lifeless> SpamapS: apt-get install does that pretty well
[03:35] <SpamapS> lifeless: and if your machine sucks.. some of the items can spin up in EC2/private cloud for better scale out during the tests
[03:35] <lifeless> SpamapS: thats massively slower again
[03:36] <SpamapS> lifeless: unless the nodes are already spun up and you're just slapping a container on top
[03:36] <lifeless> SpamapS: I repeat, I can see how you *can* do this, I don't see why I'd *want LP devs doing it*
[03:37] <SpamapS> Interesting. launchpad is I'd say one of the more complex deployments.. and yet its still desirable to have a machine that can do it all.
[03:37] <SpamapS> I would not think that.
[03:37] <lifeless> SpamapS: there are, IMO, three deployment scenarios
[03:37] <lifeless> SpamapS: local dev fast-iteration
[03:37] <SpamapS> I would think you'd want the deploy and test to be the same thing w/ different HW. :-P
[03:38] <lifeless> SpamapS: entire-complex iteration (CI tool, etc)
[03:38] <lifeless> SpamapS: and production (qastaging, edge, production itself etc)
[03:38] <SpamapS> Its sort of the opposite of what many web dev shops are doing now w/ Vagrant and other tools
[03:38] <lifeless> SpamapS: These three things have vastly different confidentiality, performance, *accessability* and debugging characteristics
[03:39] <SpamapS> But that may be because their code is not as easily configurable as LP
[03:39] <lifeless> SpamapS: an example; we don't want a breakpoint in a production server instance, ever.
[03:39] <lifeless> SpamapS: but during a test iteration and fiddle - almost certainly.
[03:40] <SpamapS> Right.. I guess my point is that in order to make this work.. I assume you have a set of static "local" configs that have to be maintained in parallel w/ the production ones, yes?
[03:40] <lifeless> SpamapS: I don't see having to rsync to an lxc container to add that being a benefit for devs, and I don't see any substantial risk for the inner loop being done on local stuff
[03:40] <SpamapS> rsync? no no no
[03:40] <SpamapS> snapshot, boot
[03:41] <lifeless> SpamapS: how many seconds will that take?
[03:41] <lifeless> SpamapS: best case
[03:41] <SpamapS> depends on your snapshotting tech
[03:41] <lifeless> say snapshotting is free
[03:41] <SpamapS> if you're on LVM or btrfs, closer to 0 than 1
[03:41] <lifeless> SpamapS: and you're snapshotting your /home ...
[03:41] <SpamapS> Should work similar to schroot, but with network separation.
[03:42] <SpamapS> Yes
[03:42] <lifeless> so, a pretty big burden, IME
[03:42] <lifeless> its unnessary for 99% of things developers have to do
[03:42] <SpamapS> snapshotting doubles write load
[03:42] <lifeless> SpamapS: I'm talking complexity
[03:43] <SpamapS> I have looked into another machine type of just "local"
[03:43] <lifeless> SpamapS: LP is massively complex already. Adding another complex stack to it makes it even harder for devs to get their head around it
[03:43] <SpamapS> where it just deploys the formula locally
[03:43] <lifeless> SpamapS: I don't want devs having to even consider all the multi-machine stuff
[03:43] <lifeless> it should be behind APIs that they call, fin.
[03:44] <lifeless> SpamapS: remember too that we're really not in a cloud deployment story: 300GB db's, multi terabyte archives - these things don't play all that nicely with VM's :)
[03:45] <SpamapS> lifeless: the idea is that they'd just model the way their service relationships.. the api would be "ensemble deploy-stack launchpad" and their environment would define how to do snapshots or chroots, or lxc's or spawn ec2's ...
[03:45] <SpamapS> err
[03:45] <SpamapS> the way their services relate
[03:45] <lifeless> SpamapS: right, or we just don't do anything.
[03:45] <lifeless> SpamapS: and developers continue on happy as a happy thing.
[03:45] <SpamapS> happy but breaking stuff because they test on maverick and deploy onto lucid?
[03:45] <lifeless> SpamapS: they don't
[03:46] <lifeless> SpamapS: we break stuff, but not for that reason
[03:46]  * SpamapS puts on old school non-devops sysadmin hat
[03:46] <lifeless> SpamapS: your story here is a non-issue for LP.
[03:46] <SpamapS> you break stuff because thats what developers do!
[03:46]  * SpamapS takes hat off
[03:46] <SpamapS> lifeless: you guys have already solved this
[03:46] <lifeless> SpamapS: we break stuff because we don't have automation where we should; and bringing up new instances of scale-out components is one of those areas.
[03:47] <SpamapS> lifeless: so indeed, LP is not in need of better service management, nor is it targetted at the cloud...
[03:47] <lifeless> I'm very interested in ensemble for the scalable components
[03:47] <lifeless> just not for developers (they don't need to scale a deployment locally, not by default), and probably not for CI (though it is a possibility)
[03:48] <SpamapS> lifeless: the lifecycle ensemble is trying to ease is    develop single instance web app on laptop -> deploy onto ec2 -> scale out on ec2 -> get successful and try to bring in house
[03:49] <lifeless> SpamapS: I don't get it :)
[03:49] <lifeless> SpamapS: I don't understand the link between the first step and the rest
[03:49] <SpamapS> lifeless: thats the lifecycle we hear about when talking to startups who are stuck on ec2
[03:50] <SpamapS> lifeless: you don't get it, because you write modular, configurable code...
[03:50] <SpamapS> lifeless: the ruby kids aren't really as mature in those skills.. ;)
[03:52] <SpamapS> lifeless: also you have *the entire Canonical Platform Team* at your beck and call  ;)
[03:52] <lifeless> SpamapS: I wish, oh what I could do ...
[03:53] <lifeless> SpamapS: seriously though, yes, we write code thats designed to be deployable
[03:53] <SpamapS> Yeah, so the ruby kids write stuff and want to test the deploy fast.. faster than they can by deploying to EC2
[03:53] <SpamapS> right now virtualbox+vagrant gives them that but its still a whole giant VM to spawn
[03:55] <SpamapS> lifeless: but I see that the basic assumptions of developers salivating over containers may be flawed when dealing with mature code bases. :)
[03:59] <ScottK> So what do containers give salivating code monkeys they don't get from rails/capistrano (I think that's what they're using in one project I'm on right now)
[04:01] <SpamapS> ScottK: most of the existing systems deal very well with static systems... not so well with cloud instances.
[04:01] <ScottK> OK.
[04:01] <SpamapS> ScottK: the exception to that is Chef, which has a nice model for handling those.. but scales *horribly* because it is such a beast.
[04:03] <ScottK> Because if I could figure a way to partition this data usefully into small chunks so I could distribute the load accross something big, then I might want a code deploying in the cloud thingy.
[04:03] <SpamapS> ScottK: of course, OpsCode will be happy to scale the server up for you, at a price. :)
[04:03] <ScottK> Sure.
[04:03] <SpamapS> ScottK: unfortunately I'd say ensemble is just getting past "cool toy" phase into "useful tool for the brave"
[04:04] <SpamapS> ScottK: specifically, the data snapshotting stuff is handled by libhandwaving0
[04:04] <ScottK> Maybe it'll be mature by the time we get the data split model working.
[04:04] <SpamapS> which AFAIK, is not thread safe.
[04:05] <SpamapS> To be fair to chef.. when I say horribly.. I mean it can't scale past the 30 - 40 nodes of one type, without sacrificing response time.
[04:05] <SpamapS> So.. if ensemble isn't there for you... chef might be a decent choice.
[04:07] <ScottK> Maybe.
[04:28] <Saturn2888> Ubuntu Server isn't installing properly from the USB. I have this problem every time I install via USB. Does anyone know why it doesn't detect it from the installer? It says "Insert CD-ROM" or something and gives this error: http://launchpadlibrarian.net/47554702/ubuntu-1004-install-fail.gif
[06:30] <MTecknology> Is there any easy way to see what apps have stuff in swap?
[06:54] <jmarsden> MTecknology: run top, press O p and then Enter      # sorts processes by swap usage
[06:56] <MTecknology> ooh.. thanks :)
[07:02] <jmarsden> MTecknology: You're welcome
[07:55] <a7ndrew> I'm trying to enable acls on my root ext4 filesystem. Asides from adding acl to the options in fstab and doing a mount -a, is there anything else necessary?
[07:56] <a7ndrew> i'm wondering if a reboot is going to be necessary
[07:59] <soren> a7ndrew: mount -a won't apply new mount options. "mount -o remount,acl /" should do the trick, though.
[08:04] <a7ndrew> soren: It did the trick, thanks!
[12:21] <guampa> anyone using make-kpkg? dunno why it started appending its own alphanumeric suffix to compiled kernel and debs, besides my --append-to-version string
[12:41] <Daviey> guampa, You might have more success asking in #ubuntu-kernel
[12:42] <guampa> sounds a good deal, i'll go there. thx
[12:43]  * zul shakes his fist
[12:46] <zul> Daviey: you know that rebuild you did for the server seed at one point in the natty release cycle? can you do it again so we can see the multiarch changes that we need to do
[12:46] <zul> its not like you have anything else to do ;)
[12:47] <Daviey> zul, Yeah.. that isn't a bad idea... not something i can do until probably tomorrow now...
[12:48] <zul> Daviey: thats fine
[12:48] <Daviey> I think it took about 24 hours to build the packages we are subscribed to last time.
[12:51] <soren> php5 needs fixing, but I assume you've seen that.
[13:07] <RoAkSoAx> morning all
[13:11] <raphink> hello
[13:48] <smoser> kirkland, fyi, uec-image is still broken with console-setup 1.57ubuntu17
[13:51] <soren> smoser: How is it broken?
[13:52] <soren> smoser: I'm using the uec-images for some tests, so I wonder if it's something I'd see.
[13:52] <smoser> they dont build
[13:52] <soren> Ah.
[13:52] <soren> Great :)
[13:52] <smoser> vm-builder copies console-setup from /etc/ -> target
[13:52] <soren> That won't affect my builds.
[13:52] <smoser> which was questionable at best
[13:52] <smoser> and now doesn't work
[13:52] <soren> Yeah, that's dreadful.
[13:53] <smoser> i'm glad you agree :)
[13:53] <soren> I always hated that.
[13:53] <smoser> you have a solutoin ?
[13:53] <smoser> because it looks like i have to come up with one
[13:53] <smoser> RSN
[13:53] <soren> In concepts, not in code. I think.
[13:53] <soren> I think I started working on it in days past, but I don't think I got anywhere useful.
[13:54] <soren> Easy solution: Don't attempt to copy any settings from the host. Just do whatever the installer does.
[13:54] <kirkland> smoser: can i see a full log?
[13:55] <soren> smoser: The solution that would be in line with the original intent of that code would be to attempt to derive the current settings in the host and then apply the same settings in the guest without assuming they share the exact same mechanics for setting these things.
[13:55] <smoser> kirkland, http://uec-images.ubuntu.com/server/natty/20110322/
[13:55] <soren> smoser: So, e.g. try to figure out which locale is used on the host and then apply that locale in the guest in a manner compatible with whatever version is being built.
[13:56] <soren> ...but just skipping that step altogether should get you close to a usable solution.
[14:00] <smoser> i really didn't want another thing that was MUST_FIX
[14:00] <smoser> but i do thikn that kirkland's console-setup work is really nice
[14:01] <smoser> other than the whole making my life suck thing :)
[14:01] <kirkland> smoser: :-/  sorry dude
[14:01] <kirkland> smoser: i'm happy to help, if i can
[14:02] <kirkland> smoser: why isn't console-setup installable in the chroot?
[14:02] <smoser> soren, ^
[14:03]  * smoser says "that was before my time" or "i just work here" 
[14:03] <kirkland> smoser: does it require hardware access or something?
[14:03] <smoser> i suspect that is the case.
[14:03] <kirkland> smoser: note that there's also console-setup-mini
[14:08] <soren> smoser, kirkland: Sorry, guys, I'm really tied up right now.
[14:08] <soren> smoser: I think removing the bit of vmbuilder that copies stuff from the host is the right thing to do.
[14:10] <smoser> thanks soren . i'll poke at it later today.
[14:10] <RoAkSoAx> kirkland: howdy1! forgot to remind you yesterday about the binaries for python-ethtool :)
[14:12] <iggi> Can anyone help me troubleshoot a UFW issue? I had a working web server prior to reboot, but after reboot ufw is blocking connections even when it says apache2 is in the exceptions
[14:13] <RoAkSoAx> zu/win 11
[14:19] <RoAkSoAx> lamont: hi there! Do you think it would be worth to consider bug #739297
[14:28] <RoAkSoAx> SpamapS: can you take care of bug #738219 and bug #738209 when you have the change please?
[14:34] <emindmedia> Hello, anyone here know how to recover a username?
[14:35] <twister004> hi guys... how can view a samba folder structure in ubuntu server?
[14:35] <twister004> I have samba shares on another ubuntu machine, which I wanna see on my server(VNC session)
[15:16] <jamespage> hggdh: ping
[15:16] <kirkland> RoAkSoAx: k, getting them now
[15:18] <WoLfLoonie> Hello everyone. I'm having a bit of an issue figuring out how to change the screen mode on console.. after installing 10.10 it became really small.. Tried searching a bit for that, as I remember there should be a choice on a reconfigure option or something.. but I only got results about editing grub2..
[15:19] <RoAkSoAx> kirkland: thank you ;)
[15:20] <kirkland> RoAkSoAx: done
[15:21] <RoAkSoAx> kirkland: thanks ;)
[15:21] <RoAkSoAx> chuck_: ^^
[15:21] <zul> hmmm?
[15:21] <RoAkSoAx> zul: python-ethtool
[15:21] <RoAkSoAx> :)
[15:21] <zul> RoAkSoAx: what about it?
[15:22] <RoAkSoAx> zul: binaries accepted
[15:22] <zul> RoAkSoAx: oh sweet
[15:23] <hggdh> jamespage, I am all ears
[15:23] <hggdh> but a bit deaf
[15:24] <jamespage> hggdh: right then; I've finally managed to get time to setup another ec2 instance and migrate the current Hudson configuration + archive etc.. onto it.
[15:24] <hggdh> jamespage, super!
[15:24] <jamespage> its got lots of storage for jenkins - 20GB on a separate ebs volume
[15:24] <hggdh> wow!
[15:24] <jamespage> Any chance you can re-point one of your slaves so I can test it works OK?
[15:25] <jamespage> URL is : http://ec2-184-72-204-166.compute-1.amazonaws.com/
[15:25] <hggdh> of course, getting it done now
[15:29] <RoAkSoAx> SpamapS: https://wiki.ubuntu.com/ClusterStack/Natty#HA MySQL
[15:29] <jamespage> hggdh: thanks
[15:31] <WoLfLoonie> hmm. I think I've fixed it through grub.. GRUB_CMDLINE_LINUX="nomodeset"
[15:38] <rynop> I'm using ubuntu 10.10 64bit daily build ami in ec2. When i ssh in i see there is a newer "server image" release. Is there a way to upgrade to the newer server image without having to rebuild my EBS?
[15:38] <hggdh> jamespage, running, the output states connected, but jenkins does not seem to see it
[15:38] <hggdh> jamespage, this is slave nhanga
[15:39] <jamespage> hggdh: hmmm - might be that it had an old hudson slave.jar cached
[15:41] <jamespage> hggdh: try shutting it down, rm /var/lib/ubuntu-server-iso-testing/.slave.jar (I think) and start up again.
[15:47] <knoxy> Hi. I've a HP DL120 with Smart Array B110i. I create RAID 1+0 with 4 HD's and Ubuntu installer cannot detect this logical volume. How can I do to fix it? I'm trying with Ubuntu Server 10.10 amd64
[15:48] <jamespage> hggdh: I think thats the issue - just tried connecting one from my locality and it worked OK.
[15:54] <hggdh> jamespage, nhanga is up
[15:54] <hggdh> that was it indeed
[15:54] <jamespage> hggdh: thought it might be \o/
[15:54] <hggdh> :-)
[15:55] <hggdh> jamespage, worth the effort to add in the postinst for next update?
[15:56] <jamespage> hggdh: I'll have to take a look and see why that happened - I think the python script might be doing sometihing odd.
[15:56] <jamespage> It should always download a new slave.jar - just in case the master has been upgraded.
[15:57] <jamespage> hggdh: PS do you like the Font?
[15:57] <hggdh> jamespage, yes, I do. And I think I have seen it somewhere else ;-)
[15:57] <SpamapS> RoAkSoAx: would be cool to write a plugin for pacemaker that could just manipulate upstart via dbus
[15:58] <RoAkSoAx> SpamapS: there is one but it is not recommended :)
[15:58] <SpamapS> RoAkSoAx: we should maybe put some dev time into that... ;)
[15:59] <RoAkSoAx> SpamapS: let's discuss it post meeting
[15:59] <SpamapS> indeed
[16:32] <lamont> RoAkSoAx: it's on my list to get that done, I rather expect that it's a big enough change that I do not plan to propose it for natty
[16:32] <lamont> RoAkSoAx: and frankly, it'll probably be a week before I actually force myself to take time to deal with it
[16:33] <RoAkSoAx> lamont: ok :), and I was thinking the same (big change for natty)
[16:52] <RoAkSoAx> SpamapS: so anyways... the reason why upstart is not the best for a cluster is becuase upstart only knows that the application is running or not, but doesn;t really know if the application is doing what's supposed to be doing
[16:53] <RoAkSoAx> SpamapS: for i.e. upstart will now that mysql is running, but for the cluster that's not enough
[16:54] <RoAkSoAx> SpamapS: because, I might want to know if I can pull certain info from a table in mysql, or even access a table, allowing the cluster to determine "yeah, it is up and running, and serving X type of requests, so I can confirm that I should bring up the other services"
[16:54] <SpamapS> RoAkSoAx: sure it does. Thats what post-start is for.
[16:55] <RoAkSoAx> SpamapS: in turn, the OCF Resource Agents, provide the capability to determine whether the resource is running, but additionally, you can also have diferent dephs of monitoring
[16:55] <SpamapS> RoAkSoAx: anything beyond that is the realm of monitoring
[16:55] <SpamapS> RoAkSoAx: so the plugin, it seems, is missing the ability to define a monitoring script.
[16:56] <RoAkSoAx> SpamapS: i guess that can be done
[16:56] <RoAkSoAx> SpamapS: but afaik, kirkland tried to add support to monitor a daemon in upstart by adding scripts but was rejected
[16:57] <SpamapS> yes, its been deemed out of scope
[16:57] <SpamapS> thats no reason to give up on it though. :)
[16:58] <SpamapS> there are two things that I agree to disagree w/ regarding upstart. pidfiles aren't always evil, and there should be a status stanza.
[16:59] <RoAkSoAx> SpamapS: right, but from HA Clustering point of view, upstart should just only tell us if the dameon is running or not
[16:59] <RoAkSoAx> however, as I said, that's sometimes not enough
[17:00] <RoAkSoAx> SpamapS: that's why LSB initi scripts are not recommended either
[17:01] <RoAkSoAx> SpamapS: but anyways, upstream themselves have rejected one of the patches that are necessary for it to work
[17:01] <RoAkSoAx> and have given recommendations to the creator to fix it
[17:02] <RoAkSoAx> but yet, it hasn't; been fixed
[17:02] <RoAkSoAx> SpamapS: and upstream are even considering in dropping the upstart support
[17:02] <RoAkSoAx> SpamapS: becuase as per their standards, OCF Resource Agents should be used to monitor resources
[17:03] <RoAkSoAx> SpamapS: the whole problem was that when Ubuntu switched to upstart, everything was broken becuase all those services with no OCF RA's were treated as LSB initi scripts, which in turn created a problem
[17:03] <RoAkSoAx> because of no return codes
[17:06] <RoAkSoAx> SpamapS: http://www.gossamer-threads.com/lists/linuxha/dev/68379?search_string=possible%20deadlock%20in%20lrmd;#68379
[17:07] <SpamapS> RoAkSoAx: good enough > perfect but not ready yet
[17:09] <RoAkSoAx> SpamapS: right but the real sisue with upstart is that it leaves open IPC sockets
[17:09] <RoAkSoAx> s/with upstart/with the upstart plugin
[17:11] <SpamapS> IPC? whats it doing w/ IPC?
[17:12] <RoAkSoAx> SpamapS: corosync is the communication layer that pacemaker uses between nodes of the cluster
[17:13] <RoAkSoAx> SpamapS: so, using the upstart plugin as Resource Agents causes corosync to leave open IPC sockets
[17:15] <SpamapS> RoAkSoAx: ah sounds like its unrelated to usptart
[17:16] <RoAkSoAx> SpamapS: not with upstart, but with the upstart pluging
[17:16] <SpamapS> right
[17:22] <RoAkSoAx> SpamapS: yeah... so anyways, mregardless of weather upstrart is fully supported in the HA cluster stack, OCF RA's are still recommened for service monitoring
[17:27] <bcessa> hi there, I'm trying to install oracle-xe on AMD64 ubuntu server, I added this to package sources "deb http://oss.oracle.com/debian unstable main non-free" but I get this error message as soon as I try to run apt-get upgrade: W: Failed to fetch http://oss.oracle.com/debian/dists/unstable/Release  Unable to find expected entry  main/binary-amd64/Packages in Meta-index file (malformed Release file?)
[17:32] <Mathuin> I have questions about Ubuntu and EC2.  Is this the right place to ask them?
[17:33] <pankaj_sharma> yes
[17:33] <Mathuin> I currently do some computations on my home server that take about 12 hours on a 8G 4x3GHz machine.  If I moved that to the cloud, could I do it in less clock time and not pay an arm and a leg.
[17:33] <Mathuin> The computations are all in Python.
[17:41] <bcessa> so the problem is my server is AMD64 and the packages in the repo are i386, stupid question but is there a way around this or a manual installation is required?
[17:48] <thesheff17> Mathuin: is your python multi threaded?
[17:49] <asmod4n> oh hi, installed apparmor and dovecot, now everytime my email client connects to it apparmor spams dmesg with ALLOWED messages... how to disable that?
[17:51] <asmod4n> apparmor seems to send everything it does to the kernel log.. is that the default behavior?
[17:53] <jdstrand> asmod4n: you installed apparmor-profiles. apparmor-profiles has profiles for dovecot that are in complain mode, which log things like you are seeing
[17:53] <asmod4n> so i just have to set them to enforce?
[17:53] <jdstrand> asmod4n: see https://help.ubuntu.com/community/AppArmor#Disable%20one%20profile on how to disable it. alternatively, you could enable the profile and have it in enforcing mode
[17:53] <jdstrand> asmod4n: yes
[17:57] <thesheff17> bcessa: you can enable 32 bit libs: apt-get install ia32-libs and you should be able to run any 32 bit program.
[18:00] <bcessa> thesheff17: thnx for your answer, I already have the package installed, how ever when I add the oracle repo to the sources.list I keep getting the following error when trying to run apt-get update: W: Failed to fetch http://oss.oracle.com/debian/dists/unstable/Release  Unable to find expected entry  main/binary-amd64/Packages in Meta-index file (malformed Release file?)
[18:00] <bcessa> don't really know how to "force" the system into using those packages or if it will work at all
[18:03] <thesheff17> bcessa: what are you trying to accomplish?  what programming language with oracle?
[18:04] <bcessa> I just wanna install the oracle server for an existing application, no programming involved on my end
[18:06] <thesheff17> bcessa: hmm...there has to be 64 bit for oracle.
[18:07] <bcessa> that will be really nice, I was trygin with this and get the error I mentioned above https://help.ubuntu.com/community/Oracle
[18:07] <bcessa> there's also some tutorials for manual installations but don't think is the best way to go, if there's no alternative I'll do it, I'm just trying to find another/cleaner way
[18:10] <thesheff17> bcessa: well the deb files are here. http://oss.oracle.com/debian/dists/unstable/non-free/binary-i386/ you could download them and do a dpkg -i packageName...it should find all the required packages...use with caution though ;)
[18:11] <thesheff17> bcessa: also check this out http://littlebrain.org/2008/05/12/how-to-install-oracle-xe-in-ubuntu-64-bit/
[18:11] <bcessa> I'll give it a go and let you know I my machine just blow LOL :)
[18:11] <thesheff17> bcessa: basically forcing the packages to install
[19:00] <RoAkSoAx> zul: could you please take care of bug #619712. I just decided to patch it
[19:01] <bcessa> thesheff17: just finish the installation and everything seems to be working just fine, thanx a lot ;)
[19:02] <bcessa> I'll configure and test the installation further ;)
[19:02] <zul> RoAkSoAx: yep as soon as im off this call
[19:02] <thesheff17> bcessa: no problem I have used oracle xe w/ ubuntu before and it is weird they don't have 64 bit support.
[19:03]  * RoyK wonders why people use oracle xe when there are so much good open alternatives, such as postgresql
[19:04] <RoAkSoAx> zul: sure take your time. thanks ;)
[19:07] <Mathuin> The installer for 10.10 desktop didn't allow me to build a RAID.  Stinky.  What's the proper way to do that nowadays?
[19:08] <thesheff17> Mathuin: you have to use the alternative cd
[19:08] <Mathuin> Oh for pete's sake.  There's a lost hour.  Sigh. :-)
[19:09] <Mathuin> Thanks for the tip.
[19:17] <RoAkSoAx> kirkland: ping
[19:18] <kirkland> RoAkSoAx: pong
[19:18] <RoAkSoAx> kirkland: quick question. When someone contributes to PowerNap with let's say a new Monitor, should the copyright be assigned to canonical/us, or just the authors?
[19:19] <kirkland> RoAkSoAx: PowerNap is not governed by the Canonical CA
[19:23] <Aison> evening
[19:24] <Aison> i'm searching a movie database or organizer web software. Eg. where ppl can search for a movie etc...
[19:24] <Aison> I would like to install it on my ubuntu server (with about 1000 movies stored on it)
[19:32] <RoyK> Aison: there are several - google for it
[19:33] <Aison> I did
[19:33] <RoyK> one of my first hits was http://www.techsupportalert.com/best-free-dvd-collection-software.htm
[19:36] <Aison> :)
[19:43] <_1094kms> I just created a lvm volume from scratch (new hd). When I run vgchange -ay I get an errormessage :  http://paste.ubuntu.com/583948/  is this something to worry about?
[19:59] <patdk-wk> that isn't good
[20:00] <_1094kms> got my answer - I had old volumes still in the cache and vgchange -ay checks them all while they were not existing anymore.
[20:09] <rynop> anyone used tmpfs to back their web app sessions before? I'm using PHP and looking to use tmpfs instead of db/file backed.  Wondering if there is some pitfall i'm not thinking bout.
[20:11] <RoyK> rynop: not really - should work, except load balancing won't be easy across servers if they can't share the sessions
[20:13] <rynop> RoyK, I have the session affinity covered cuz i'm using amazon elastic load balancer - so I think i'm only in trouble if my server goes down - which in taht case i don't care bout the session data :)
[20:14] <RoyK> ok
[20:14] <rynop> RoyK, in the reading i've done, tmpfs falls over to swap space if it ever uses up the allocated mem space right?
[20:15] <rynop> and if i allocate 200meg of mem to tmpfs, and i only use 100meg, the extra 100meg is available to the system right?
[20:17] <RoyK> rynop: think so - iirc it allocates 50% of RAM
[20:17] <RoyK> rynop: and the allocation is dynamic, not like old ramdrives that used a specific amount
[20:19] <rynop> RoyK, cool thx.
[20:20] <RoyK> meaning if you use 30kB, it'll use 30kB of your precious RAM
[20:20] <RoyK> if on a low-memory system, remember to increase swappiness, so that the kernel can swap out stuff not in used and rather use the memory for something useful
[20:21] <RoyK> echo vm.swappiness >> /etc/sysctl.conf && sysctl -p
[20:21] <RoyK> erm - wrong
[20:21] <rynop> isnt it: /bin/echo “1″ > /proc/sys/vm/swappiness
[20:21] <RoyK> echo vm.swappiness=100 >> /etc/sysctl.conf && sysctl -p
[20:21] <rynop> oh
[20:21] <RoyK> rynop: swappiness is set to 0-100 - setting it to 1 will make linux hardly swap at all until the shit really hits the fan
[20:22] <RoyK> which is not what you want
[20:22] <RoyK> echo 100 > /proc/sys/vm/swappiness will set it, but not store it over a reboot
[20:22] <rynop> RoyK, dont i only want to do swappyiness of 100 for my tmpfs drive? not the other processes using memory?
[20:23] <RoyK> swappiness is system global
[20:23] <RoyK> but it usually makes sense to set it high
[20:23] <RoyK> that is, unless you're doing something like realtime linux, but then you don't want swap at all
[20:24] <rynop> what does swappiness default to? i dont currently have any 'swap' strings in my /etc/sysctl.conf file
[20:24] <RoyK> 60
[20:25] <rynop> I see, so setting it to 100 wont rip out a session in memory (in tmpfs)? it woudl only rip it out if its not in use (duno how it knows that but i'll trust u)
[20:25] <rynop> why is the default not 100? seems like there is no downside.
[20:27] <RoyK> rynop: google for it - there have been a lot of discussions on it. I have a bunch of servers all set to 100, and it works well. Worst case is if you have some lesser-used process that gets swapped out and it takes a second or two to get it back
[20:27] <RoyK> that is - a whole process won't be swapped out - just pages not in use
[20:28] <rynop> RoyK, ah ok thx.
[20:28] <RoyK> erm - pages not used very recently, that is
[21:25] <JoePrro> hi im new to linux and irc. I was hoping to find someone with some experience to share
[21:26] <JoePrro> Im am setting up a ubuntu server...it has six scsi drives 2 36.4 gbs and 4 146gb drives.  the 36s are raid 1 and the 146s are raid 5 on a serveraid 7k card.  i want to install ubuntu server on raid one array with home folder and mysql on raid 5.  what is the best practice for this?
[21:30] <guampa> JoePrro: put / in the raid 1, and two partitions for /var and /home in the raid 5 ... maybe / could be split in more partitions if you wish, for /, /boot and /tmp  f.e
[21:35] <JoePrro> guampa: I am actually still in the partioner stage of the ubuntu setup.  it see the the arrays as two disks scsi3(0,0,0) & scsi3(0,1,0) Do I install ubuntu and swap on 36 gig or do I create a software raid5 with those two disks
[21:36] <JoePrro> i unsure how software raid works or is needed with a hardware raid setup
[21:36] <guampa> if you don't have a controller for those two and have to go software you have to decide between 0 and 1 afaik
[21:37] <JoePrro> i do have a controller... its an ibm xseries with serveraid 7k card
[21:37] <JoePrro> so no special setup. no need for software raid setup?
[21:38] <guampa> if you can make the two arrays hw based, go for it. hasn't always got the same flexibility as sw but everyone and his mom keeps saying that they aren't even near in terms of performanve
[21:39] <guampa> performance
[21:40] <JoePrro> so harware is the answer...huh? So just install ubuntu & swap on raid1 and then configure folders to reside on raid 5
[21:41] <guampa> ditto. a common choice for fs's would be XFS for the raid 5 areas, and say, EXT4 for the two-disk array
[21:41] <BbluE> Hello everyone.
[21:42] <BbluE> Everyone afk?
[21:43] <JoePrro> makes sense
[21:44] <BbluE> I guess.. :(
[21:44] <JoePrro> does the xfs need to be primary or logical
[21:46] <BbluE> I think Primary, but don't quote me.
[21:46] <guampa> since you will only boot linux it more or less doesn't matter if the parts are primary or logical, or even if you don't use partitioning and go directly to LVM over raid
[21:47] <guampa> the schemes make important to pay atention to what OS loader you use in some situations, but i believe GRUB2 should work in almost every scenario
[21:53] <JoePrro> great thank you...so I have configured sda 34.9gb ext4 and #5 sdb 1.5 GB swap and now i need 2 partions on xfs volume one for var and one for home...right?
[21:54] <JoePrro> or just one large partition on xfs volume
[21:57] <guampa> yes, you can put them in the same partition too. i don't know much about the specifics of your server, you can decide based on the amount and type of data you'll be putting in /home and /var
[21:59] <JoePrro> its a basic lamp server...running i use ispconfig3 phpmyadmin and joomla
[22:01] <JoePrro> i have it up right now on a dell vostro with just a 500 gig sata drive. I am trying to put redudancy on now before i go live with my site
[22:01] <BbluE> I'm trying to set up a basic lamp server too.. it's destroying me from the inside out! lol.
[22:01] <JoePrro> lol
[22:01] <JoePrro> i feel ya man
[22:02] <guampa> if it's going to be small then go with the big partition, no need to add worries then :)
[22:03] <JoePrro> yes small....sortof...its a membership site so mysql will grow large
[22:04] <BbluE> I started this morning, trying to make my dang .htaccess work... Google, is finding me more questions like mine, with no answers.. I have a few different Ubuntu PDF books, and I think those are just talking me around my issue.. but, not directly through it... etc, etc.
[22:05] <BbluE> I think he means to just go with the one partition, instead of adding multiple... and keeping it simple
[22:05] <JoePrro> BbluE: Howtoforge got me up and going really good stuff there
[22:06] <BbluE> Sweet, I'll go check that out. Thank you for the lead.
[22:07] <JoePrro> thanks so much for help guampa...I think i was making this a bigger monster than what it is
[22:10] <guampa> sorry was afk, yes setting up servers are a great adventure, even more the first ones
[22:10] <guampa> s/are/is
[22:10] <guampa> still tuning my english :P
[22:10] <guampa> anyway glad to be of help
[22:11]  * RoyK pats his 100TB OpenIndiana boxes
[22:11] <BbluE> Lol, :D
[23:00] <RoAkSoAx> kirkland: do you think bug #735452 is worth uploading for natty?
[23:03] <hackeron> hey, I try to do: hal-find-by-capability --capability video4linux - in maverick it showed a list of v4l devices - in natty it shows nothing - is there an equivalent command?
[23:45] <PoorNewb23> hi there
[23:46] <PoorNewb23> Why use Ubuntu-server, and not any other flavour?
[23:46] <PoorNewb23> (it's serious question)
[23:46] <PoorNewb23> I couldn't find any concrete answer in help/faq and general webpage.
[23:47] <ruben23> hi guys does ekiga have g711 codec-ulaw and alaw
[23:54] <hackeron> PoorNewb23: Ubuntu server is pretty barebone, you don't have a graphical user interface, it's just typing commands - it's suitable when you don't plan to ever plug a keyboard and mouse to the PC/server after it's set up
[23:54] <hackeron> PoorNewb23: and naturally no monitor either
[23:55] <hackeron> ruben23: I believe so, you can use ekiga as an asterisk client when asterisk is configured to use ulaw/alaw
[23:59] <ruben23>  hackeron: you have idea how do i do conference call with ekiga on asterisk..?