[00:16] New bug: #1088720 in facter (main) "facter virtual runs lspci unconditionally" [Undecided,New] https://launchpad.net/bugs/1088720 [01:50] Is there a definitive list of all the kernel boot parameters for Ubuntu Server? And yes, pointing me to source code is an acceptable answer. [01:51] After doing a mountain of PXE installs the last few days I figure it's time to learn more about boot options and preseed files [02:33] hey I'm looking into preseed files and think I got the hang with setting up raid1 and lvm, but now I wanna encrypt the lvm how can I do that? [02:33] I couldn't find any documentation... === NomadJim_ is now known as NomadJim [03:48] sarnold, Hellowwwwzz :) [03:52] hey I'm using preseed files and wanted to set the hostname, but I still get prompted when specifying get_hostname. I have dhcp enabled, but I sort of think dhcp should be enabled and the hostname should be set anyway... === Ursinha is now known as Ursinha-afk [05:01] is it possible to change the order apache modules get loaded? [05:03] I'm trying to use rewrite & vhost_alias, but rewrite appears to functioning based on /var/www/ being the document root, rather than the automagically created /var/www/%1/public document root. === Gallomimia_ is now known as Gallomimia [05:14] nevermind, I just changed the redirect in my .htaccess, which I don't like but I'll just live with until I can find a way to make Apache generate a more accurate document root [05:26] looks like the issue with vhost_alias was fixed in apache 2.4, I'll just mess with my .htaccess files until apache 2.4 is integrated into ubuntu-server [05:26] (I realize that is not a small task) === cpg is now known as cpg|away === ibiris|afk is now known as ibiris [06:53] lot of mysql updates today, anyone had problems after upgrade?? === cpg|away is now known as cpg === soren_ is now known as soren === RaringOne is now known as IdleOne === smb` is now known as smb === tspxx is now known as tspx [10:26] zul or anyone familiar with openstack packages: why does nova-api-metadata breaks: nova-api? === mcclurmc_away is now known as mcclurmc [10:38] Adri2000: nova-api provides all the various api frontends. [10:39] Adri2000: nova-api-metadata provides only the metadata frontend, so the two would conflict (each trying to grab the metadata port) if installed at the same time [10:52] soren: ok. then do you know what's the use case of having a separate -metadata package? is -metadata supposed to be installed on all compute nodes? (in all/some network configurations?) [11:20] Adri2000, co-incidentally I was just looking into this [11:29] Adri2000, soren: I really think that should be some sort of Conflicts/Provides rather than a Breaks bearing in mind what that relationship is trying todo [11:29] i.e. prevent installation of nova-api + any of the other subpackages. [11:47] jamespage: How come? [11:48] Adri2000: Well, you just have the option of running the metadata server independently of the api servers. Whichever way you choose to do so is up to you. [11:48] soren, well Breaks is not quite as strong as Conflicts; a Conflicts relation ensures that the packages cannot be installed/unpacked at the same time [11:48] Adri2000: On each compute host is a good idea, but not required by any means. [11:49] jamespage: And what does Breaks prevent and why don't you think that's sufficient? [11:50] jamespage: There are no files in conflict. Only ports. [11:52] soren: I understand that; [11:53] soren: say I install nova-api and then I install nova-api-os-compute; I think this will force a deconfigure of nova-api prior to install on nova-api-os-compute [11:53] but not a removal... [11:53] which is what I think the behaviour should be [11:54] jamespage: Why? [11:54] I mean... dpkg doesn't just leave it there unconfigured. It'll get removed. [11:54] soren, will it? [11:54] i think it will just not let it be reconfigured [11:55] * jamespage goes to try this [11:56] soren, OK - I'm wrong [11:56] it works just fine [11:57] * jamespage scopes that [11:57] *wrong in the context of "i think it will just not let it be reconfigured" [11:58] Yeah. Installed, but not configured is a transitional state. [11:59] Hrm... Terminology fail there. [11:59] "Unpacked, but not configured" is what I meant. [12:00] "Installed" specifically means that it's unpacked and configured. [12:00] brb [12:00] soren, well whatever - its sufficient to ensure that the nova-api and nova-api-* packages don't get installed together === niemeyer_ is now known as niemeyer [12:13] jamespage: indeed [12:19] soren, I still think Conflicts is the correct approach even though Breaks does the job [12:19] I normally use Breaks with Replaces (both versioned) when move files between packages. [13:36] hallyn: re netcf, could it be the testsuite? [13:47] Hi. I just upgraded mysql-server (5.1) on my 10.04 server and it wont start because of missing /etc/init/mysql.conf ... that seems owned by mysql-server-5.1 according to dpkg -S and dpkg -L [13:47] A reinstall of mysql-server-5.1 doesn't fix it. Also strangely "debsums mysql-server-5.1" doesn't even list that file [13:47] What am I doing wrong here? === frankban_ is now known as frankban === cpg is now known as cpg|away [13:50] VSpike: could try to purge the package, note it might remove databases too so make backup first [13:50] so take a copy of /var/lib/mysql before purge [13:51] that sounds a bit harsh [14:01] Is there a way I can extract that specific file from the package and copy it into place? [14:03] Yep, that works :) [14:08] Hi anyone knows an extreamly simple server management tool? something like a script on each server, that tells it's ip address and other infos to a master server? === niemeyer_ is now known as niemeyer [14:14] sebastian__: as a one-time thing? [14:14] maybe a cron job every hour or so [14:15] should work on linux & mac os x [14:16] sebastian__: There are a few ssh clients that will do simultaeneous connections to remote servers that you might be able to script to acheive that. mssh, mussh, pssh, taktuk, clusterssh seem to fit the bill (from apt-cache search multi ssh) [14:17] Of course, there might be a better tool that does that out-of-the-box, but I don't know of any off the top of my head. [14:20] well, to execute somthing wih mssh for example, I need a list of all servers. I would need my server to add itself to the master server [14:28] jdstrand: no, wasn't that (testsuite not runnable on buildds, netlink)) [14:30] I thought there might have been progress on that front, but I'll take your word for it [14:40] stgraber: jinkeys that guido guy is hard to parse [14:40] jdstrand: i'll kick off a test in ppa to make sure [14:43] hallyn: hehe, yeah. I think I vaguely understood what he wants, but I'm not planning to spend any time actually making that happen myself ;) [14:46] jamespage: ill trade you one [14:46] zul, sure [14:46] jamespage: https://code.launchpad.net/~zulcss/keystone/buh-bye-dbconfig/+merge/139215 [14:50] kees: stgraber: ok, i see libseccomp 1.0.1 is in raring-proposed. i'll fix up lxc to make it work with it, but are there other packages depending on libseccomp which will need updates? [14:51] hallyn: according to apt-cache, no [14:51] k [14:52] and reverse-depends seems to confirm that only lxc build-depends on it (reverse-depends -b libseccomp-dev) [14:52] hallyn: ping netdev right? [14:53] zul: ? [14:53] hallyn: the sync you asked me to do last night [14:54] zul: netcf, yeah === nveitch is now known as evilnickveitch [14:56] zul: (i'm doing a test run in ppa with tests enabled, but i expect those to still fail in buildds) [15:19] stgraber: the proper seccomp fix for lxc will require some autoconf. do you object to an autoconf-free fix in raring package, expecting the nicer fix to come with the 0.9.0-alpha1 merge? [15:19] i'm fine doing it the other way, only downside is libseccomp 1.0.1 sits in -proposed longer [15:20] * hallyn is torn [15:20] will do quick fix first locally to make sure it even works :) [15:21] oh, haha [15:21] actually all of the seccomp code is autoconf-free in the package. never mind. [15:23] ok :) [15:27] zul: see mp - one minor problem [15:28] jamespage: k thanks === jcastro__ is now known as jcastro_ === acidflash_ is now known as acidflash [15:41] hello masters! [15:43] i have a problem on my server....from time to time, my server start a 'dhclient' and i loose my ip fix address.... [15:43] how can i fix that? [15:43] thanks! [15:47] im trying to disaBle virbr0. i dont need it plus its flooding dmesg--->http://fpaste.org/7kZO/ [16:13] jamespage: fixed up the comments === matsubara is now known as matsubara-lunch [16:19] stgraber: so do i just upload a new lxc it'll build against the new seccomp, and both automatically get promoted? or is there a different workflow? === phunyguy_ is now known as phunyguy_t430s [16:23] ivoks, do you fancy finishing off bug 671065? [16:23] Launchpad bug 671065 in dovecot "deliver broken because dovecot.conf uses !include_try" [Medium,In progress] https://launchpad.net/bugs/671065 [16:26] jamespage: i might [16:27] ivoks, I'll barter with you for a final review of the mail-delivery charm [16:27] :-) [16:28] heh [16:28] i need to take a vacation to actually finish the work on ubuntu stuff :) [16:32] ivoks, lol [16:32] hallyn, ping. [16:33] random curiosity. i think that i can attach disk images to lxc containers now ? [16:34] zul, sorry - more comments === Ursinha-afk is now known as Ursinha [16:35] hallyn: that should just work, yes [16:36] jamespage: argh [16:47] Stoopid question. How safe is it to upgrade an 11 server up to 12 that is running on mdadm? [16:47] I dont exactly have a place to backup my data [16:47] hey is there ANY documentation about how to configure raid1 + encrypted lvm at all for preseed files? [16:48] scalability-junk: it's possible to do raid1+lvm and ecrypted+lvm via preseed, easily. I am trying to make all three, but failing so far. [16:49] xnox, yeah we talked a few days ago and I fail too. [16:49] thought doing a manual config and then make the configs into a preseed file, but that fails :P [16:50] only tab tab etc. choices... not really helping. [16:50] yeah.... [16:52] xnox, give me a shot if you figure it out :P [16:55] CaptSammy: be very careful [16:55] CaptSammy: first off, make sure you're using 12.04.1, 12.04.0 shipped with a particularly nasty mdraid bug [16:55] rbasak, no -proposed cloud-imags. what i'd likely to do test this stuff is boot existing image, add proposed, update && upgrade, rm -Rf /var/lib/cloud /var/log/cloud && reboot [16:55] OK [16:56] ACK! Good to know [16:56] CaptSammy: save all pertinent data about the array itself, /proc/mdstat, mdadm -D /dev/mdX, mdadm -E /dev/DEVICES, save it all into a notepad file somewhere, in case it melts down [16:56] If I am using the automated upgrade, how do you know what ver you are going to? [16:57] if you're jumping to the most recent 12.04.1 you'll be using mdadm 3.2.5 [16:57] mayube I dont NEED to upgrade :) [16:57] unless its absolutely necessary, i really wouldn't, ask yourself why you need to upgrade to begin with [16:58] Gotta keep up with the Joneses! [16:58] xnox, that's my raid1 lvm setup not sure yet if it works... now trying to figure out how to merge it with my encrypted lvm setup... [16:58] http://pastebin.com/KsbYyhrY [16:58] I was going tro install newznab and figured it might be a good time to upgrade the system at the same time [16:59] 11.04/11.10 are showing it's age, i think if you are simply prepared, you'll be fine [16:59] if, after the upgrade, your array won't assemble, just don't panic, come to irc or email the linux-raid list for assistance [17:00] maybe I should grab an external 2T drive and backup first [17:01] The array is my big fear, I am not raid savvy [17:01] and I got the entire machine booting and running on that mdadm array [17:02] xnox, refined one: http://pastebin.com/BXGr6bZp I'll test it now and wish me luck. [17:03] raid isn't a backup, so yes, you should have a backup [17:03] Whats a good app for backing up my server to an external nightly? [17:03] cron + rsync [17:03] drives are cheap enough, might as well do things right [17:04] * RoyK uses bacula [17:04] bacula might be more "user friendly"? [17:04] rsync won't keep old versions [17:05] bacula is a bit hard to setup at first, but working with it (with bat etc) is rather user-friendly [17:05] * qhartman uses dirvish [17:06] all the benefits of rsync, but with historical snapshots and a really robust expiration policy manager [17:06] dirvish is a good alternative too [17:06] is it command line friendly? [17:06] afaik dirvish is only commandline ;) [17:06] Dirvish is really command-line only [17:07] no gui at all [17:07] that's totally not true, you need to script rsync to properly handle versioning. [17:07] My system is of course headless and not GUI [17:07] yeah, I use dirvish to backup a server in Denver to a server in Eugene every day [17:07] have been for years, and it's been totally bullet proof [17:07] CaptSammy: my server is headless as well, but running bat from it over ssh is quite easy with x11 forwarding [17:08] * RoyK usually have x libs installed on his servers [17:08] some software is easier to use with a GUI, after all [17:11] some software have a small webserver for local administration, usually listening to localhost only - running firefox http://localhost:something is way more userfriendly than using lynx or similar text mode browsers, and it doesn't make the system more insecure [17:12] does dirvish or bacula have web front ends? [17:13] dirvish does not. You basically build a config that defines your backup policy, set a cronjob to run the backup, and that's it [17:13] to get files or whatever, you just browse the file system [17:13] use normal commands to copy files back or whatever you need [17:15] CaptSammy: bacula has webacula, but like the rest of bacula, it's got a bit steep learning curve to get it installed - works well when it's installed, though [17:15] I'll throw in another suggestion: rsnapshot [17:15] that is, you have webacula and a few others [17:15] jrib, I also use rsnapshot in a few places, and it also works well. [17:15] Really any of the mature tools are arguably good choices [17:16] you just need to decide which one fits your use case best [17:16] it's all a discussion between whatever fits or what you like [17:16] I've gotten used to how bacula works, so I use that... [17:16] indeed [17:16] On bacula.org they show the web front end, looks nice [17:17] url? [17:17] should be relativily easy since I am only backing the one machine up? [17:17] http://www.bacula.org/images/bweb1.png [17:18] CaptSammy, it really feels like you're overthinking this. Pick one, try it out, if you don;t like it, try something else. [17:18] tru [17:18] CaptSammy: setting up bacula is a bit hard anyway, since it's an enterprise system, really - it makes little difference in setting up bacula for one or fifty servers, somehow [17:18] CaptSammy: with bacula, you have to setup retention policies and virtual tapes, not quite like the others mentioned here [17:18] Bacula probably is overkill, RoyK is not kidding when he says it's got a learning curve. I've looked at it a couple times, and decided against it every time,. [17:19] qhartman: I originally set it up for a five-site backup with some 200 machines [17:19] the tape thing is the killer for me, it's designed to work with legacy tape library semantics, which is extra complexity I don't need, and am not used to. [17:19] qhartman: now I'm using it for 7 machines or so, single site [17:19] RoyK, right, and for that environment it probably made a lot of sense, and since it's still owkring, no point in changing. [17:19] but then, I've learned how to work with it :P [17:20] bummer, I like the web control [17:21] CaptSammy: spend an hour with the bacula config - it's a good howto on their sites - and bweb or webacula or something else can do fine [17:21] CaptSammy: there's also #bacula [17:22] on their *site* even [17:22] * RoyK just uses bat === matsubara-lunch is now known as matsubara [17:22] that's the bacula X interface [17:31] Thanks for all the insight [17:37] I am going to install a 3T drive and backup before trying an upgrade [17:38] Amazon has a 3T drive for 150 bucks [17:38] I remember when a ten meg drive was around six hundred bucks [17:39] and thought "who could ever fill this thing"? [17:40] CaptSammy: we've all thought that way ;) [17:40] CaptSammy: keep in mind that that single 3TB drive may just die one day without giving you a warning [17:41] Is there a way to spin down a drive when not used for an active backup? [17:41] CaptSammy: better get some cheap 1TB drives and setup a NAS (or a DIY-NAS with an old PC) [17:41] CaptSammy: yes, you can do that, but normally the best a drive can have, is a stable atmosphere with stable temperature [17:42] If that 3T dies the only way I could lose data is if my Raid5 dies atr the same time though [17:42] man hdparm [17:42] It exists a ubuntu dnssec doc? [17:42] there's an "auto spindown" there iirc [17:44] Thanks again! [17:46] any of you run newznab? [17:48] * RoyK doesn't === shadeslayer is now known as buildslayer === mcclurmc is now known as mcclurmc_away === ibiris is now known as ibiris|afk [19:43] im surpised no one has asked for samba 4 yet [19:44] zul: its been in Ubuntu/Debian for a long time as a beta/alpha [19:45] zul: why would I want to ask for it, my current samba server work just fine :) [19:45] SpamapS: yeah but its no longer beta/alpha [19:45] I know [19:46] it looks like the RC is in raring [19:46] I'm sure the final will ship [19:54] hallyn: https://bugs.launchpad.net/ubuntu/+source/ifupdown/+bug/1065684 is quite a pain to fix ;) [19:54] Launchpad bug 1065684 in ifupdown "network-interface-security.conf needs to go away when the interface does." [Medium,Confirmed] [19:54] hallyn: I managed to decrease the number of -security devices by half at least [19:55] hallyn: the problem we still end up with is that upstart doesn't get any event when a device is moved to another namespace... [19:56] hallyn: so on a veth pair, we can only properly flush the host device but not the container device [19:57] hallyn: I think we already discussed this but I'm seeing a few options to make the problem completely go away, not sure if you have any preference [19:57] hallyn: 1) Get the kernel to emit net-device-removed + net-device-added when a device is moved between netns [19:57] hallyn: 2) Have lxc store the original name of the device and instead of destroying it in the netns, move it back to the host with its original name, then destroy it [19:59] hallyn: 3) Introduce some kind of net-device-moved kind of event we could use to stop the various jobs (though that's a bit tricky as we'd need to detect the case where a device is moved back to the host) [20:02] hallyn: 1 or 2 would fix a whole bunch of bugs we're getting at the moment where various piece of software try to do state tracking of net devices and fail with containers [20:03] hallyn: I actually believe that's the source of the memory leak I've seen in network-manager here as it's getting net-device-added events but no net-device-removed, it's still storing data as if it still existed [20:03] hallyn: which considering I created around a thousand veth devices since yesterday, wastes a fair bit of memory :P [20:07] jinkeys - now that's a backlog [20:08] stgraber: there is a kernel patch for sending event when dev is moved to another ns [20:08] it's in net-next [20:08] we could ask rtg to pull it into raring [20:09] in other words, option (1) is done in net-next [20:10] hallyn: can you point me to the commit and I'll e-mail ubuntu-kernel about it for inclusion so I can close that bug for good ;) [20:10] that's a problem :) [20:10] cause i can never find that tree. lemme check linux-next [20:11] stgraber: http://git.kernel.org/?p=linux/kernel/git/next/linux-next.git;a=commitdiff;h=4e66ae2ea371cf431283e2cb95480eb860432856 [20:11] stgraber: thanks for pursuing that :) [20:12] * hallyn goes to finish this pair of libvirt bugs finally [20:13] hallyn: thanks! I was just going to tell you to that net-next is owned by David Miller but you were too quick to find the commit in -next ;) [20:14] hah - i know it's *owned* by dmiller, but i never know where to find it. it's not on kernel.org lasti i knew [20:14] so i every time i want to do something based on net-next... i end up stalling [20:14] :) [20:18] stgraber: feh. i sent seccomp patches to the mailing list from hallyn.com. lxc-devel is holding them. i forgot the list admin password (or dlezcano changed it). so i'm just going to push to staging... [20:19] stgraber: if you want to look first, it's github.com/hallyn/lxc.git #staging.dec11.1.seccomp.2 [20:20] hallyn: can you hold on posting to staging? I'd actually prefer to have what's currently in staging be released as 0.9.0~alpha1 (so it matches my pull request) [20:20] hallyn: I'm trying to get Daniel to do the release ASAP so we can start pushing stuff to it again [20:21] stgraber: actually this might ought to go to alpha, bc otherwise it will FTBFS on debian [20:21] (I should have branched to 0.9.alpha1 on github and use that for the pull-request, but only thought of it after sending the e-mail...) [20:21] hmm, fair enough, push it then :) [20:22] stgraber: ok [20:24] pushed. thx. i likely won't touch the tree again until next year. [20:26] stgraber: heh 'hello *kernel*people*' [20:26] Here can i take advantage to have any help about dnssec on ubuntu? [20:27] hallyn: :) [20:28] stgraber: no ':)' to the last paragraph though. Y I K E S [20:28] hallyn: just took a look at the seccomp patch, looks good. I assume you did a test build with/without seccomp to make sure nothing blows up? [20:31] yup, and with old+new api === cpg|away is now known as cpg [20:44] anybody can help me on dnssec? [21:17] <_KaszpiR_> question [21:17] <_KaszpiR_> i've got server on some old motherboard with e6550, its on bios [21:17] <_KaszpiR_> now I'd like to move the ssytem to moterboard with UEFI - any ideas what to do/avoid etc? === SJrX is now known as SJr [21:56] how do I check if network is the bottleneck? I have a tacacs+ daemon which intermittenly do not respond to router [21:56] running on ubuntu-server [21:57] netstat -ni shows TX-ERR 4304824 and TX-OK 1228922546 [21:57] no RX-ERR === niemeyer_ is now known as niemeyer === matsubara is now known as matsubara-afk [22:21] I'm having a weird issue - I have the resolvconf package installed, but nothing i put in /etc/resolvconf/resolv.conf.d/tail or /base has any effect [22:35] <_KaszpiR_> axisys depend on the card, on the wifi this can be ok, but on wired link you got serious issue [22:47] hi [22:47] I need help with http://232aae397ea4bcf2.paste.se/ [22:48] can you help me? [23:14] jdstrand: fwiw the netcf tests still fail on buildds :( [23:20] zul: jamespage: smoser: do you have any comments on the documentation/note in http://people.canonical.com/~serge/qemu-kvm-sysctl.debdiff ? [23:31] jdstrand: as the patches have been acked upstream, i'm going to push lp:~serge-hallyn/ubuntu/raring/libvirt/libvirt-hugepages to the archive. === acidflash_ is now known as acidflash [23:38] (that is, if i don't see an objection before tonight :) [23:38] * hallyn bbl [23:40] hallyn: ack, I saw that. I clanced at them and they seemed reasonable [23:40] Even as Ubuntu's desktop flavor annoys me more, I keep finding new things to like about the server version [23:40] hallyn: thanks for that :)