[00:32] hi [00:32] how do i change file names from console? [00:32] in Linux [01:25] can anyone clue me in on /usr/share/X11/locale? [03:04] normally if I wanted to define a dead key I would do it in the Compose page for the locale, but spanish for example doesn't have a locale defined here, instead using teh default. If I wanted to flesh out spanish with my own custom locale so I could manipulate the dead keys, what would I need to build in that directory? [04:24] Anyone know how to boot lucid alternate from usb? alternate REALLY wants a cdrom. Posted solutions for other variants doesn't work. [04:30] blistov: I know people have had success using usb-creator-kde from Lucid with at least the server CD (which is the same installer as the alternate CD). [04:33] ScottK, any idea if the alternate cd is going get fixed in the future? [04:33] I don't think it's the alternate CD that's broken. [04:35] It boots from USB, but refuses to read the install media from the boot media (unless the boot media happens to be a cdrom ). [04:36] Every other distribution of Linux (including Ubuntu) is able to correctly read install media from ... any physical media. [04:37] This is a huge problem for server installations. [04:37] I'm not sure our company even owns a server with a cdroom. [04:40] blistov, try unetbootin maybe? [04:43] johanbr, Thats what I'm using. [04:43] Again, I can boot the image, but the installer doesn't even give you ann option to install from any media other than a cd. [04:43] so you didn't try usb-creator? [04:45] no. [04:45] trying now, but I can't imaging its going to make changes to the actual installer... [04:46] then I would recommend that, that's the official method for putting images on usb media [04:59] blistov: I helped someone with this exact problem a few days about and usb-creator-kde worked for them. [05:09] ScottK, and johanbr, That did it. [05:09] Thanks a tonne guys. [05:09] blistov: Please go complain to unetbootin about not working for this situation. [05:09] you're welcome [05:09] Yes, you're welcome. [05:13] :) [05:41] yes usb creator works for server installs , its how i do all the server installs that arent PXE here === Amaranth is now known as c === c is now known as Amaranth [06:15] blistov: you are wrong, period [06:15] I have installed alternate/server over the network, using pxe [06:15] no offence, but if it doesn't work, that's because you are doing sth wrong, not because sth is broken === al-maisan is now known as almaisan-away === almaisan-away is now known as al-maisan === al-maisan is now known as almaisan-away [07:15] ugh.. us.archive.ubuntu.com seems to be made up of hosts with an allergy to Los Angeles.. 512kbit? Really? :-( [07:15] * SpamapS hopes University of Oregon is happy w/ his mirroring. ;) [07:23] SpamapS: us.a.u.c is london [07:23] SpamapS: but you should be able to get more than 512kbit out of them [07:24] SpamapS: also #ubuntu-mirror for mirror stuff [07:24] sorry, -mirrors [07:26] Ugh I'm in something like 12 ubuntu- channels already.. how many must I join? ;-) [07:26] * SpamapS takes his bleary eyed sleep-needing whining where it belongs.. to bed. ;) [07:26] no, you don't have to join; was just letting you know [07:26] thats cool I may actually ask about it in there [07:44] kees: well, that's the plan, but it's apparently not that easy to do, and so we disable the script in lucid for now to prevent more damage; Chase was going to change it into an example script which you can copy to /etc/ [07:45] pitti: whoa.. I don't know if thats wrong window or major lag.. but.. what script are you referring to? [08:05] hello === ogra_ is now known as ogra === almaisan-away is now known as al-maisan [10:40] When does the debian synching occur for Maverick? E.g. when will udhcp be imported from sid? https://bugs.launchpad.net/ubuntu/+source/udhcp/+bug/566845 [10:40] Ubuntu bug 566845 in udhcp (Ubuntu) "Please update udhcp from debian" [Undecided,New] [10:42] alkisg: syncing is already happening [10:42] alkisg: that bug is equivalent to "cjwatson needs to get round to merging busybox [10:42] " [10:42] Ah, got it :) [10:42] Heh [10:42] Thank you, I'll just check periodically then. [11:49] cjwatson: as regards LP: #580763 what version shall I use for the package in maverick? [11:51] tseliot: up to you, as long as it's distinct and greater [11:51] is libc6-i686 needed or is it in the libc6 package? [11:51] tseliot: if you want to use -0ubuntu1 for maverick and -0ubuntu1~10.04 or something for lucid-proposed, that's fine too [11:51] ^^ 10.10 [11:51] gnomefreak: no longer needed and in fact no longer existing [11:51] cjwatson: thanks [11:52] we're moving to i686 across the board === amitk is now known as amitk-afk [11:52] cjwatson: ah, ok I'll use ~10.04, thanks [11:53] cjwatson: I think you'll have to reject my last upload before I can do that though [11:53] tseliot: not actually true, but I can do it anyway [11:53] (done) [11:53] cjwatson: thanks === pgraner-afk is now known as pgraner [12:30] in Lucid, Compose sequences can only consist of two characters, and may only output one character. This makes them exactly equivalent to dead keys. but many useful example ~/.XCompose files online make use of 3 combining characters (in vitro w/ 2) or output multiple keys (digraphs) [12:30] how can I get that functionality ? === oubiwann is now known as oubiwann_ === amitk-afk is now known as amitk === oubiwann_ is now known as oubiwann [13:13] any testdrive devel here? [13:13] * hyperair wonders what to make of all sabdfl's mails having a bad signature [13:14] que? [13:14] well, so says enigmail anyway =p [13:14] * xnox loves sabdfl's french =) [13:14] gpg: Signature made Monday 24,May,2010 08:02:57 PM SGT using DSA key ID D54F0847 [13:14] gpg: BAD signature from "Mark Shuttleworth " [13:15] hmm [13:15] =p [13:15] could be a MIME / T-bird thing. which mail? [13:15] the one you just sent on ayatana around 13 minutes ago [13:15] but yeah, it could be [13:16] i think enigmail has some issues with html mail [13:27] and - why does compose-c-c give me a č - everywhere in gnome apps but no where in qt apps? its not defined in my system's compose files: http://pastebin.org/274106 === JFo-swap is now known as JFo [14:22] and - why does compose-c-c give me a č - everywhere in gnome apps but no where in qt apps? its not defined in my system's compose files: http://pastebin.org/274106 === sconklin-gone is now known as sconklin [15:08] kirkland: ping, do you happen to be around atm? [15:08] g: yes [15:09] kirkland: have you had a chance to look at LP#571093 (libvirtd eating memory over time) the last couple of days? [15:10] kirkland: I think I might be able to supply a rough patch to prevent it (trying to build it now) [15:10] g: we know that it exists, and there are a handful of upstream memory leak changes in upstream libvirt git [15:10] g: i haven't had a changes to zero in on it yet [15:10] g: if you have a patch, i'd love to see it, will help get it sponsored [15:11] kirkland: AFAIK upstream havn't changed this function in 0.8, I'm trying to prevent it by making sure we free the memory if we are returning non 0 [15:12] (or any of the related functions for that matter) [15:12] g: oh? [15:13] kirkland: I think this is a new bug due to udev rules in Debian/Ubuntu, I couldn't reproduce it under Fedora, but the udev rules there are totally different [15:14] g: oh, interesting, thanks for narrowing that down [15:14] in Lucid, Compose sequences can only consist of two characters, and may only output one character. This makes them exactly equivalent to dead keys. but many useful example ~/.XCompose files online make use of 3 combining characters (in vitro w/ 2) or output multiple keys (digraphs) [15:15] kirkland: I'm Nigel Jones in the bug btw [15:15] how do I get that functionality [15:15] hmmmm great, certainly reduces the effect of the leak [15:15] g: thanks [15:15] g: but does not prevent it? [15:15] kirkland: valgrind showed two major leaks [15:15] I'll take a look at the second now [15:16] I wonder if it's leaving a device entry around somewhere [15:17] it's only about 8k RES during a multipath -F/multipath -v4 though so it has reduced [15:18] blue_anna: it would seem nobody here knows. Perhaps you should try mailing the ubuntu-x list? [15:18] g: gotcha === blue_anna is now known as blue_annacj === blue_annacj is now known as blue_anna [15:19] cjwatson: thanks, I guess I'll try that too === dendro-afk is now known as dendrobates [15:36] kirkland: okay, I think I got rid of the second memory leak, I'll attach the patch to the launchpad with my explanation for the second fix too === pretto_ is now known as Pretto [15:45] g: thanks [15:47] kirkland: just posted my patch [15:48] * kirkland checks [15:49] kirkland: RES memory use appears to be stable now except for the odd increase/decrease but it no longer corresponds to when I run multipath [15:50] g: patch looks perfectly reasonable to me [15:51] g: could you take the to-do of sending this patch to the upstream list, while I work on getting the SRU together? [15:51] kirkland: sure [15:52] if I can remember my RHBZ password that is :P [15:54] g: post the libvir list [15:54] oh okay, even better :) [15:55] I guess they'll want it in git format iirc? [15:56] jdstrand: around? [15:56] jdstrand: G just posted a simple patch that purportedly fixes https://bugs.edge.launchpad.net/ubuntu/+source/multipath-tools/+bug/571093 [15:56] Ubuntu bug 571093 in multipath-tools (Ubuntu) "multipath + libvirtd eats away more memory over time" [Medium,Triaged] [15:57] jdstrand: i'm just verifying that, and will upload a package to lucid-proposed [15:57] jdstrand: just wanted to give you a chance for feedback first ;-) [16:07] g: can you give me a few specific instructions on reproducing the leak? [16:09] kirkland: okay, well the easiest way I found was that w/ multipath having no configuration multipath -F followed by multipath -v4 would generate the required udev add/change/remove calls [16:13] kirkland: an examplehttp://paste.ubuntu.com/438911/ [16:15] SpamapS: I was replying to kees, not to you [16:18] kirkland: any method of triggering the add/remove udev calls should reproduce the issue btw [16:19] g: cool, i'm on it [16:25] g: hmm, I'm running this, without seeing a leak: [16:25] for i in $(seq 1 10000); do sudo multipath -v4 >/dev/null; sudo multipath -F >/dev/null; free; done [16:26] kirkland: do you see anything in udevadm monitor? [16:27] g: hmm, no [16:27] kirkland: oh, I found running 'top -p (pidof libvirtd)' was the best way of keeping track of it [16:30] g: yeah, nothing, yet [16:30] g: no events in the monitor [16:30] kirkland: what ouput do you get from 'multipath -F; multipath -v4' [16:30] *output [16:36] pitti: I understand you weren't talking directly to me. I was just curious what sort of script was doing damage. [16:36] SpamapS: the pm-utils-powersave-policy one for spining down SATA drives [16:37] pitti: interesting [16:38] g: http://paste.ubuntu.com/438920/ [16:41] kirkland: hmmm, the only thing different is 'mapp already present' [16:42] kirkland: if you do multipathd -k and at the prompt run 'show config' it's empty? [16:42] g: correct [16:43] the only thing I can think of is at first, I had an issue where udev was in a weird state [16:43] (when I was first reproducing it) [16:44] I could reproduce it, then udev would act up and stop producing the add/remove calls [16:45] ah ha [16:45] kirkland: yeah, thats the issue [16:46] kirkland: got an entry in /dev/mapper by any chance? [16:46] for your HD that is [16:46] g: http://pastebin.ubuntu.com/438923/ [16:47] kirkland: bingo, I think we may have struck on a race condition in multipath? :P [16:48] g: hrm, interesting ... [16:48] because I managed to get to the same point with 'while true; do sudo multipath -F; sudo multipath -v4; done; [16:48] kirkland: tbh, when I got to that point I just rebooted [16:49] and reproduced with: top -p (pid of libvirtd) in one terminal, and manually running 'multipath -F; multipath -v4' in another terminal and watching RES going up [16:49] g: i'm running your while true, no leak [16:49] 1047 root 20 0 188m 7192 2960 S 0 0.2 0:00.37 libvirtd [16:49] g: i'm pegged there [16:49] g: 7192 constant [16:50] kirkland: if that map already present message is still there, I'd expect it to peg there [16:52] g: so how do I try to re-trigger the race? reboot and let udev redect my hard disk? [16:52] kirkland: I'm just trying to work out what multipath is looking for regarding that issue [17:01] kirkland: ah ha! [17:01] kirkland: without a reboot: dmsetup ls, if it's there with (252, 0) do a 'dmsetup remove ' [17:02] kirkland: that fixed my multipath [17:02] so it's actually a dmsetup/multipath race === gnomefreak76 is now known as gnomefreak === RoAk is now known as andreserl [17:10] g: hmm, okay, so this is a bit outside my expertise [17:10] g: what's dmsetup remove going to do? [17:11] kirkland: remove the map that multipath created but couldn't remove, if like me, the HDD is mounted traditionally (like /dev/sda1 etc) then it won't effect mounted partitions [17:11] dmsetup = device mapper, and just creates fancy references, mainly used for multipath & LVM [17:11] g: i'm mounted by UUID [17:11] kirkland: yeah, UUIDs won't be effected [17:12] g: and after I do this, I'll reproduce the leak? [17:12] yep should do, but do it slowly with just: multipath -F; multipath -v4 [17:12] and wait a sec before doing it again [17:13] I think the problme with the do while loops is it tries to flush before the map is created [17:14] g: okay, now i have activity in udev admin [17:14] g: and yeah, now i see a memory leak [17:14] good and the RES memory usage should be going up for libvirt [17:14] g: correct [17:15] kirkland: great :) [17:15] g: okay, so I've reproduced this; can you post an update to the bug with your race condition analysis? [17:15] with the patch the starting RES should be lower than without (because it's fixed for startup) and then also it shouldn't go up during normal usage [17:15] kirkland: actually, the race condition is seperate, it's really a new bug [17:16] which I'll create [17:16] g: i see your post to the libvir list, thanks ;-) [17:17] kirkland: was just waiting till I was certain that you'd be able to reproduce the issue :) [17:17] g: heh, okay, i just need to document how to reproduce it for the SRU [17:18] kirkland: no problem [17:26] doko_: Did you plan on uploading your last Ubuntu twisted fix to Debian so we can sync it? I see some builds failing for lack of an update and a sync would be nicer than a merge.... [17:27] kirkland: btw, for the race condition: https://bugs.launchpad.net/ubuntu/+source/multipath-tools/+bug/585027 [17:27] Ubuntu bug 585027 in multipath-tools (Ubuntu) "Race condition with dmsetup causes 'map already present' messages" [Undecided,New] [17:27] kirkland: anything else I need to do, or can I just leave it with you? [17:28] g: just one sec, review my update [17:28] kirkland: sure, let me know when it's up [17:30] g: http://pastebin.ubuntu.com/438949/ [17:32] kirkland: for IMPACT: I'd remove the system loses a udev race on boot bit [17:32] the race condition that we experienced just impacted our ability to hit the bug [17:32] the impact is really any system w/ libvirtd running while udev add calls are occuring [17:33] kirkland: to avoid the race condition in the while true loop, maybe add a sleep 1 or something like that to let udev/dmsetup/multipath to settle [17:33] g: okay === deryck is now known as deryck[lunch] [17:34] kirkland: but yeah, that sounds about right [17:35] (I'd put the sleep 1 bit after multipath -v4 [17:37] g: thanks, uploaded to lucid-proposed [17:42] kirkland: cool, thanks for looking at it, I think I might look at some of the libvirt bugs sometime too, that one just got me really interested though :) [17:42] g: cheers, thank you; *great* job tracking it down [17:42] g: the libvirt bugs could certainly use a hug from someone with an interest in fixing some of them ;-) [17:43] kirkland: yeah [17:44] kirkland: must say, valgrind was the biggest help === amitk is now known as amitk-afk [17:46] kirkland: well, have a good day! :) [17:48] kirkland: oh I guess I should also forward this to the debian.org BTS so they can apply the patch there as well [17:48] g: that would be good, yeah [17:48] kirkland: if you want, I can do that in the morning [17:48] g: nice, you got an ack on the upstream patch too, great [17:48] oh, haven't seen that yet [17:49] kirkland: oh thats great news [17:51] Does ubuntu have a web front end to commits? [17:53] I've mostly got my compose issue worked out now [17:54] the main thing that's got me confused is gnome-terminal and firefox don't accept many-key compose sequences, and won't output more than one character -- it only works for definitions that are like deadkeys [17:54] everything else, xterm, opera, kde apps .. they all work with like for example compose+b+t+w = "btw" [17:54] or "by the way" I was being quick :) === oubiwann is now known as oubiwann_ === al-maisan is now known as almaisan-away === oubiwann_ is now known as oubiwann [17:56] https://help.ubuntu.com/community/ComposeKey - I was following this guide [17:58] Hit compose, release it and then hit the rest. [17:58] both ways, it works or it doesnt. like compose+t, m works. so does compose, t, m [17:58] but compose, b, t, w does not -- unless I'm using anything other than gtk [18:00] I have explicitly set GTK_IM_MODULE like it says to in the guide [18:00] ion: not sure what's up with it === deryck[lunch] is now known as deryck [18:23] ion: you still here? === dendrobates is now known as dendro-afk === dendro-afk is now known as dendrobates === amitk-afk is now known as amitk === dendrobates is now known as dendro-afk === bjarkef is now known as thefakebjarkef === thefakebjarkef is now known as bjarkef [19:36] the gtk people tell me that xim is not the preferred im because it is buggy. I am getting one of those bugs with my particular set up. what is the suggested lucid alternative to xim for custom compose keys ? https://help.ubuntu.com/community/ComposeKey -- this says xim but like I said that's known not to work === dendro-afk is now known as dendrobates === radoe_ is now known as radoe [19:50] do you know how to get Xcompose working with ibus? there are some unanswered posts on ubuntuforums about that but no solutions === xfg is now known as zul [20:50] blue_anna: I believe ibus is the recommendation [20:52] persia: yeah I heard that now, thanks === emma_ is now known as emma === dendrobates is now known as dendro-afk === dendro-afk is now known as dendrobates [22:59] sudo service statd start -- fails with Warning: Fake initctl called, doing nothing. Any ideas?