[00:08] [0x40f239] malloc(10) = 0x1959fe0 [00:08] [0x40661c] free(NULL) = [00:08] [0x40f239] malloc(2016) = 0x1958550 [00:08] is there any way to turn this into a source code line number? [00:08] (first hex number is the instruction pointer at time of the call) [00:12] bluefoxicy: If you install the dbgsym packages for ls and such the backtrace at any point should include linenumbers, IIRC. [00:13] RAOF: that didn't come from gdb [00:13] I'm not sure how to script gdb... [00:22] Oh. That'll be a strace, right? [00:23] yeah [00:23] ltrace but [00:23] I want to extract every call to malloc(), free(), and realloc() from an application, timestamped [00:44] hey [00:44] did anyone see what I said about translations? [00:45] that everything started showing in English since I upgraded the language packs and nobody seems to *care*? [00:51] PovAddict: Where did you point this issue out, and have you filed a bug or checked if there have been any bugs filed about the issue? [00:51] I pointed it out here and mentioned the ID of the bug I issued [00:51] #316174 [00:51] er, filed [00:51] whatever [00:52] Ok as long as a bug is filed, it will be attended to. [01:54] so annoying... does *everybody* here just use the desktop in English? [01:55] unlikely [01:57] * Hobbsee blinks [01:58] PovAddict: it's less likely that people here still run hardy, fwiw. [01:59] upgrading means learning a whole new desktop environment [01:59] KDE3 -> KDE4 [02:00] this is true [02:00] hm, interesting. [02:01] as a workaround, i'd suggest you install version 1:8.04+20081124 [02:02] strange thing with the -base packages [02:03] a 20090105 version is listed but not shown as an upgrade [02:04] what's the logic to decide if a package is upgradeable? :/ [02:04] use a dist-upgrade on it [02:04] that one contains files, it appears [02:05] aptitude dist-upgrade says there is nothing to do [02:06] strange. [02:07] PovAddict: oh, and it's in proposed, btw, so may not show on packages.ubuntu.com [02:07] (it hasn't been pushed to everyone, fortunately) [02:07] hmm why did I get it? I have -proposed in my sources.list but it's set in... some way so that it's not used by default [02:07] set in... some way so that it's not used by default? [02:08] they supposedly don't show unless I do aptitude -t hardy-proposed [02:08] * Hobbsee shrugs, being relatively unfamiliar with teh internals of aptitude. I usually lock things at dpkg level, so it works everywhere, if i want to do that [02:09] it was done by editing /etc/apt/preferences [02:09] I don't understand how it works though :) [02:09] Package: * [02:09] Pin: release a=hardy-proposed [02:09] Pin-Priority: 400 [02:10] anyway, I just downgraded the four packages i had upgraded before [02:10] * Hobbsee wonders if aptitude even follows apt's preferences file [02:10] should [02:11] well it has worked before :) [02:11] I also have hardy-backports there, and I'm sure they don't show unless I "-t" ask for it [02:11] aptitude -t hardy-backports shows two dozen upgradeable packages right now (that don't show otherwise) [02:12] oops, I meant -proposed in that last message [02:12] backports shows even more :) [02:25] hi i have a question and i didnt get an answer it seems on 64 bit intrepid the /usr/lib/libGl.so link to the libGl.so.180.11 isnt set [02:26] its set in the lib32 [02:26] should i add it manually [02:26] ghostcube: didn't get an answer where? [02:26] #ubuntu #kubuntu [02:27] ghostcube: Considered checking whether a bug has been filed about this? If there isn't one, I suggest you file one. [02:27] i looked and i didnt find anything useful [02:27] ghostcube: Is this problem causing applications to malfunction/not function at all? [02:28] i cant compile compiz it claims missing libGl [02:28] so i think this could cause problems [02:28] ghostcube: You need to install the development files for the nvidia libraries. [02:28] sorry why did I get nvidia mixed up in all of this? :S [02:28] hmm then it removes the libgl1-mesa-dev and the libgl1-common-dev [02:29] ghostcube: is there any reason why you need to compile compiz? [02:29] iam just supporting in channel so i need the newset versionmostly from git [02:29] don't mind me and what I said rei nvidia. [02:29] Ok, I suggest you run "sudo apt-get build-dep compiz" which will install everything you need. [02:29] yeah but this wont fix the prob its just an missing symlink [02:30] as it seems [02:30] ghostcube: How do you know it won't fix it? Have you tried it? [02:31] try what ? it worked fine i just tried to compile from git [02:31] and i get the missing liGl error [02:32] i need the git version of compiz the master not the apt-get source builds [02:34] do you know what build-dep does? [02:34] it cathces all dependencies for the build === Zhenech_ is now known as Zhenech [02:37] i dont want to stress i am just wondering if the symlink is missing [02:37] thats all :) === hyperair1 is now known as hyperair [02:40] ghostcube: No, its in a -dev package somewhere. I am not sure which one because I'd need to check what package the file belongs to. [02:40] i checked it with apt-file [02:40] it belongs to libgl1-mesa-dev [02:40] Its Debian/Ubuntu policy to keep those symbolic links in the development packages only, as they are not needed for normal use. [02:40] and this is installed [02:40] Ok. [02:41] and if i want to install the nvidia-glx-180-dev which contains the same file it will remove the libgl1-mesa-dev [02:41] is this the problem maybe ? [02:42] I don't know enough about those packages to comment. [02:42] me too :| [02:42] and it seems no one else in channels around iam asking since 4 hours now [02:43] noo one has any idea [02:43] nah 7 hours [02:43] boah late === bluesmoke is now known as Amaranth [07:04] good morning [07:04] That === warp10_ is now known as warp10 === tkamppeter__ is now known as tkamppeter === hyperair1 is now known as hyperair === hyperair1 is now known as hyperair [08:20] Keybuk: [08:20] Whoops [08:33] 14:57 < Keybuk> soren: I'm reverse-thinking here [08:33] 14:57 < Keybuk> we should probably just copy of all of /etc/udev/rules.d into the initramfs [08:34] 14:58 < Keybuk> though that might break things since we don't copy hardly any of /lib/udev/rules.d [08:34] 14:59 < Keybuk> or [08:34] Keybuk: I don't understand that last line. [08:34] 14:59 < Keybuk> copy anything not in /lib/udev/rules.d [08:34] 14:59 < Keybuk> and copy anything in /lib/udev/rules.d *if* it has been copied into the initramfs [08:36] * soren wonders if Keybuk is actually here at this hour [09:22] can we move imlib2 to universe? [09:22] to me it looks like all rdepends live in Universe [09:23] have you checked component-mismatches and/or germinate? [09:23] slangasek: erm, no [09:24] imlib2 doesn't show up on http://people.ubuntu.com/~ubuntu-archive/component-mismatches.txt, which means there's something else in main that (build-)depends on it [09:25] hm [09:25] http://people.ubuntu.com/~ubuntu-archive/germinate-output/ubuntu.jaunty/all shows that w3m build-depends on it [09:25] ah, good to know [09:28] thanks slangasek === ara_ is now known as ara === thekorn_ is now known as thekorn [09:37] anyone know where pitti is? [09:39] Hobbsee, germany? [09:39] "elsewhere" this week :) [09:39] directhex: I knew that, but thanks :) [09:39] Hobbsee, :p [09:40] slangasek: ah. So, who does one get poked about for a broken hardy-proposed langpack update? [09:40] me [09:40] Hobbsee: he's on holidays for the week [09:40] slangasek: consider yourself poked on https://bugs.launchpad.net/bugs/316174 then :) [09:40] Launchpad bug 316174 in language-pack-es "All translations gone" [Critical,Confirmed] [09:40] also ArneGoetje, who's the one that will have to fix it... [09:40] seb128: thank you [09:41] PovAddict: with regard to language packs, the shrinkage in size is entirely expected; the way language packs are structured is that there's an -LL-base which contains most translations, and an -LL which contains some more translations as updates to that; this lets us push out frequent langpack updates with less mirror churn [09:41] Hobbsee: hrm, surely that's because they've been merged into language-pack-es-base, since that was the purpose of the langpack update? [09:42] PovAddict: so what happened in -proposed was that the -LL-base packages were respun to save us CD space, which of course resulted in the -LL packages shrinking down to almost nothing [09:42] slangasek: ahhh, is that the point. Right [09:42] PovAddict: did you upgrade the -base packages as well? [09:42] slangasek: from the bug, looks like an insufficiently-tight dependency, though [09:43] * slangasek nods [09:44] hmm [09:44] Package: language-pack-es [09:44] Depends: language-pack-es-base [09:44] and [09:44] Package: language-pack-es-base [09:44] ArneGoetje: ^^ can you fix that so language-pack-$foo has a versioned dep on language-pack-$foo-base? [09:44] Depends: language-pack-es (>= 1:8.04+20090105), locales (>= 2.3.6) [09:44] Conflicts: language-pack-es (<< 1:8.04+20090105) [09:44] but there's a long-standing dpkg bug/lack-of-feature to the effect that Conflicts B->A are not checked on upgrade of A [09:44] well, surely it's not even that [09:45] and of course the old language-pack-es-base is still installed [09:45] right [09:47] -base still seems to be around the expected kind of size [09:47] so I don't think we've actually had major translations lossage [09:48] cjwatson: I think he said he couldn't update -base for some reason, so that would work [09:49] Hobbsee: he said that it wasn't listed as an available upgrade, and it seemed that aptitude was doing something confusing with pinning [09:49] that was it, yes [09:51] cjwatson: actually, the -base packages have each grown by 300k or more [09:52] er, 200k for -en, 400k or more for pt/es [09:52] which makes sense in terms of post-release translations [09:52] but poses interesting problems for LTS point releases [09:53] they have? [09:53] -rw-r--r-- 1 lp_publish lp_publish 3094350 Jun 17 2008 ubuntu/pool/main/l/language-pack-es-base/language-pack-es-base_8.04+20080527.2_all.deb [09:53] -rw-r--r-- 1 lp_publish lp_publish 3106094 Jan 10 12:04 ubuntu/pool/main/l/language-pack-es-base/language-pack-es-base_8.04+20090105_all.deb [09:53] -gnome-es-base has grown about 40K AFAICS [09:54] cjwatson: well, I was comparing against 8.04.0 [09:56] because there are packages from 8.04.1 that aren't in the main archive anymore, making it hard to do a full comparison, grmbl [09:56] (mutter, need to get something set up to record the .deb sizes at image build time) [09:58] that's why there's a hardlink tree archive of .1 on cocoplum [09:59] yeah, but I have the .list files on antimony :) [09:59] I guess copying those to cocoplum would be the easier route, wouldn't it [10:21] calc: ooo3 binaries accepted, enjoy :) [10:24] cjwatson, OOo3 landed in main? [10:25] cjwatson: Just FYI I managed to bootstrap rpm with its autogen last Friday, thanks [10:25] (had to fix a couple of issues but at least it got the work done) [10:33] directhex: yes [10:33] lool: ok, cool [10:37] Grrr [10:39] blkid just fixed itself automagically I hate that [10:42] Aha /etc/blkid.tab and /etc/blkid.tab.old [10:42] Too bad I already destroyed evidence... [10:50] Umm... Where does compiz store its config these days? I change stuff in gconf, but it doesn't seem to take effect. [10:51] mvo: ^ ? [10:53] soren: I had the same issue; there's also an "ini" backend [10:53] soren: it should be gconf, but sometimes it gets confused and starts using the ini interface. please use ccs and look at the backed settings there [10:53] soren: lower left corner [10:53] soren: You can access settings in a backend independent way with the python bindings, but I found addressing individual settings extremely painful [10:53] soren: I thought I had fixed this particular bug during intrepid development though :/ [10:53] (just borke when I upgraded to jaunty for instance) [10:54] I forget... Button2 is the right or middle button? [10:55] mvo: I used to set my shortcuts with a script like this one http://people.ubuntu.com/~lool/set-shortcut-config but the way I address the compiz settings broke completely; it there a better way to address settings? [10:55] s/it/is [10:55] I'm having the most annoying issue with compiz or ccsm or something. [10:56] I don't have a middle button, and from Xfce I'm used to alt-rightmousebutton resizes windows. I want to make compiz do the same, but I get bitten by a whole stack of weirdness. [10:57] Alt-RightMouseButton brings up the window menu. I never need that, so I just disable it under General in ccsm. [10:57] * lool finds it weird that blkid maintains his cache of block device's properties when we have udev and sysfs [10:58] When I disable the window menu, compiz seems to pretend that I'm holding down . [10:58] ! [10:58] soren: uhh, let me try to reproduce [10:58] To click on things, I need to hold down alt. To drag a window, I need to *not* hold down alt. [10:58] mvo: Hang on, I'll reset all my compiz settings. [10:58] lool: does the ccsm "export my settings" help you for your use-case? [10:59] I'll nuke ~/.config/compiz (which seems pretty empty anyway) [10:59] and to a gconftool-2 --recursive-unset (after grabbing a backup) [11:00] soren: there is a "reset" button in ccsm as well [11:00] (just fyi, your method will/should work too) [11:00] Let's just say that I don't trust ccsm very much at this point :) [11:00] soren: you may want to stop compiz before the recursive unset, sometimes the config system tries to be clever :) [11:01] mvo: Right, I did it while running metacity. [11:01] Ok. Here goes. [11:02] I go to the "General" place in ccsm and disable the window menu. [11:02] In the "General" tab, third option from the top. [11:02] As soon as I do that, it's like my key is inverted. [11:02] When I try to click on things, it thinks I want to move the window. [11:03] When I hold Alt down, I can click on things normally. [11:03] If I go to the "Move" plugin. [11:03] ... [11:03] It clearly says that the initiate button is Button1. No modifier. [11:03] That rates at least 5.6 on my weird-stuff-o-meter. [11:04] Anyhow, I add there.. [11:04] and my button's sanity is restored. [11:04] I go to the resize plugin.. [11:05] And tell it that I want the initiate button to be Button3. [11:05] It then tells me that the window menu action already uses this combo (even though I disabled it). [11:05] I click "Disable Window Menu" [11:06] And my button is b0rked again. [11:06] s/button/key/ [11:07] I'll try setting the window menu thing to Button9 instead (I'm hoping that won't conflict with anything). [11:07] mvo: I guess I could use export/import settings; never tried it; will see what that gives, thanks [11:07] What the... [11:08] That changed all the modifiers for "Resize window" as well?!? [11:08] Madness [11:08] I give up. [11:08] soren: let me try that here [11:09] I might be doing this all wrong. I just want Alt-RightMouseButton to resize windows. [11:11] soren: I can reproduce that here, I have a look (and/or forward to upstream) [11:13] Does someone know what upstream bug tracker e2fsprogs uses? It seems to be Debian's e2fsprogs source, but I'd like to confirm [11:14] I don't know for sure, but I know that Ted follows both Debian's and Ubuntu's bug tracker to some degree; probably pays more attention to Debian's [11:15] my impression is that he tracks e2fsprogs bugs in LP rather closely [11:16] hmm. the edubuntu pod now includes an edubuntu-desktop-kde seed that references an edubuntu-desktop-kde metapackage. [11:16] http://ext4.wiki.kernel.org/index.php/Bugs but I'm not sure whether that's only for the kernel side [11:17] lool: http://sourceforge.net/tracker/?atid=102406&group_id=2406&func=browse seems to exist too [11:17] soren: as a workaround, please try "preferences" and select the "enable integreation into the desktop environement" [11:18] (via e2fsprogs.sourceforge.net) [11:21] soren: please let me know if that works for you (it seems to do the trick for me) [11:23] cjwatson: Yeah, I checked bugzilla.kernel.org and came to the main conclusion, and the main e2fsprogs site, and was completely confused [11:24] Ok; I'll just Also affect e2fsprogs and ping tso [11:25] slangasek: Thanks for the hint [11:29] lool: hey, I see your name in http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=510673 [11:29] Debian bug 510673 in libjack-dev "xine-lib: FTBFS: /bin/sed: can't read /usr/lib/libsamplerate.la: No such file or directory" [Serious,Closed] [11:30] lool: there's a sponsorship request open to pull in the removal of the .la from Debian. Do you know why "bio2jack" is the only package that will break from that? [11:41] james_w: because bio2jack is the only reverse-dep which is also a shared lib shipping a .la? [11:44] iain@intrepid:~/packaging/bio2jack/usr/lib$ grep dependency libbio2jack.la [11:44] dependency_libs=' /usr/lib/libjack.la -lrt -lpthread -ldl /usr/lib/libsamplerate.la -lm' [11:45] yeah, I came out with a huge list of packages when I tried, but I think I used the tools wrong [11:52] I get "libbio2jack0-dev: /usr/lib/libbio2jack.la portaudio19-dev: /usr/lib/libportaudio.la portaudio19-dev: /usr/lib/libportaudiocpp.la libfluidsynth-dev: /usr/lib/libfluidsynth.la libecasound2.2-dev: /usr/lib/libecasound.la" [11:56] james_w: inspection of libportaudio.la seems to indicate that this lib doesn't depend on libsamplerate [11:56] mvo: That's already selected? [11:56] mvo: Sorry about the slow reply. compiz sent me to lunch. [11:59] seb128: I have just fixed a rather annoying bug in nautilus with a simple 3-lines patch. Can you have a look at it, please? http://bugzilla.gnome.org/show_bug.cgi?id=567479 [11:59] Gnome bug 567479 in Tabs "Clicking on a different tab doesn't update the scrollbar" [Normal,Unconfirmed] [12:01] tseliot: that's a duplicate of http://bugzilla.gnome.org/show_bug.cgi?id=542396 [12:01] Gnome bug 542396 in Tabs "ScrollBar isn't showed when open a new tab(impossible to scrolling)" [Major,New] [12:01] tseliot: could you add your patch to this one? [12:02] seb128: sure [12:02] tseliot: I'll try to get upstream to review it [12:02] thanks [12:03] seb128: ok, done [12:03] tseliot: thanks [12:03] thank you [12:21] and libiaxclient-dev with /usr/lib/libiaxclient.la [12:24] Riddell: any news on the knetworkmanager front? will we get a kde4 version? === ubott2 is now known as ubottu [12:50] asac: I'm hopeful but I believe it's not there yet [12:50] Riddell: damn. so how bad would another knetworkmanager breakage in alpha 3 be? [12:51] Riddell: i have the final 0.7 bits waiting for upload ... knetworkmanager probably will have issues though [12:53] asac: I guess we'll live with it, are they in the PPA to test? [12:53] Riddell: yes [12:54] ~network-manager [12:54] Riddell: let me know how bad it is .... my guess is a crash ;) [12:54] james_w: yes because its *.la files reference the dropped ones [12:55] Pff xine-lib is maintained in hg [12:56] james_w: It wasn't my recommendation to drop the *.la files; instead I recommended stripping the dependency_libs= line to be empty [12:57] james_w: It might affect a different set of packages in Ubuntu than in Debian though; it depends whether our packages shipping *.la files were built against other packages when these used to ship or stopped shipping *.la files [12:58] fta: pastebinit is broken in jaunty? [12:58] lool: ah, ok [12:58] fta: seems to not honour my .pastebinit.xml anymore [12:58] james_w: It might affect more than libjack-*-dev's rdeps, but if it does then these packages were lacking a dep on libjack*-dev [12:59] So it should be enough to check whether libjack*-dev rdeps ship *.la files and whether these reference libjack*-dev's *.la files; in theory, this should be done on all arches [13:01] lool: I have a list for i386, checking that rebuilds of them will be sufficient now. [13:06] soren: (sorry for my late reply, I was at lunch :) - please unselect the DE environment integratio [13:17] ? [13:19] mvo: Ooh... Looks promising! [13:22] * soren hugs mvo [13:22] mvo: Awesome! It works. [13:22] soren: thanks, I'm created a upstream bugreport and will add a (targeted for beta) ubuntu one too [13:22] soren: thanks for brining this up :) [13:24] You bet. [13:44] asac: I've just had a two-day saga of getting my 3G modem to work ... none of which was NetworkManager's fault. :-) As soon as I managed to teach the kernel and hal to detect it, NM Just Worked. Nice! [13:45] slangasek: I just updated Bug #316262 - It should be removed. I'd appreciate it if you'd have another look. [13:45] somewhat infuriatingly it's twice as fast as my ADSL connection [13:46] cjwatson: yeah congrats ;). is that a new modem or did you need to adjust an existing hal entry? [13:46] asac: newish modem, it's one of the hso family and needs nasty hacking [13:46] urgh [13:46] what was needed on kernel side? [13:46] cjwatson: ? [13:47] asac: deactivating the stupid ZeroCD thing it comes with [13:47] heh [13:47] yeah [13:47] sounds like you're familiar with it already :) [13:47] I think the plan is to fix the driver [13:47] yes, Dan Williams already sent a patch to do most of that [13:47] there is one modem class that already has the fix in the driver [13:47] ah ok cool [13:47] but it needed a bit further tweaking [13:47] I'll send something upstream shortly [13:48] shall I CC you? [13:48] sure, why not. [13:48] asac: may I ping you about the sunbird/iceowl 0.9 uploads? ;) [13:49] siretart: yes ;) [13:49] feel free [13:50] siretart: i think https://code.edge.launchpad.net/~mozillateam/sunbird/ubuntu-0.x and the iceowl one are really ready for action [13:51] i think its now a matter of upload bandwidth here ;) [13:52] ah, I see :-) [13:52] siretart: ok building sunbird sources and then will just push i guess [13:53] in case something is broken we probably can still fix it before release ;) [13:53] siretart: uploading [13:53] cool! thanks! [13:55] https://edge.launchpad.net/~ubuntu-dev/+polls [13:56] can't we have both? :-) [14:08] Keybuk: so you were talking about how git reset is missing from bzr ... the other way round, do you know how to do 'bzr revert ' in git? 'git reset --hard ' is what I'd expect but it refuses [14:09] (i.e. forget about all changes in index and working tree, but only to certain file(s)) [14:09] git checkout -- [14:09] ah, of course [14:09] thanks! [14:09] you may need -f there [14:09] yeah [14:10] fatal: git checkout: updating paths is incompatible with switching branches/forcing [14:10] maybe -- [14:11] ah, just without -f works [14:11] obvious UIs 'r' us [14:12] heh [14:12] ooh, right, -f is when you've already git add'd it [14:12] no, I actually need 'git checkout HEAD ', and it doesn't DTRT with new files [14:12] it's obvious when you see things from git's pov [14:12] I *had* already git add'd it [14:12] bah [14:13] git sucks ;) [14:13] no argument from me [14:17] asac: knetworkmanager appears to work with NM 0.7 from the PPA [14:18] Riddell: thx [14:18] asac: KDE 4 applet still doesn't though alas [14:19] Keybuk: is there a neater way, in hal, to say "match on the contents of this sysfs attribute" than writing a preprobe shell script to fish out the attribute and save it with hal-set-property? [14:19] Keybuk: the IHV-written code I have here does the latter, but I notice that nothing in hal or hal-info seems to do anything much complicated with preprobe, which leads me to suspect that I might be on an inelegant track [14:22] only in code [14:23] you'd have to modify HAL's code to add new sysfs attributes [14:23] is that preferred over adding a preprobe script, for something that's insanely device-specific? [14:23] I guess [14:24] maybe I'll just send a patch to hal-info with the preprobe script and see what David says [14:24] less effort if it's a bit ambiguous :-) [14:24] hal_util_set_string_from_file (d, "my.little.property", sysfs_path, "wibble") [14:48] slangasek: about the versioned dependencies on the language-packs: I can try. Since the langpacks are generated by script, the script would need to be ammended to accept the dependent version number as argument... [14:57] ArneGoetje: this surely is not a problem since language-pack-LL already has versioned Replaces on various other packages with the correct version numbers [14:57] and therefore must have this information already [14:57] cjwatson, you busy? [14:58] directhex: sort of, why? [14:58] more stack-full than busy if you see what I mean [14:58] debian bug 480020. can you think of any major reason not to insist to novell that they apply the included patch immediately? [14:58] Debian bug 480020 in openssh-server "openssh-server: adjusted OOM killer is inherited by all child processes" [Normal,Closed] http://bugs.debian.org/480020 [14:59] cjwatson: but not for the -base packages, right? [14:59] i'm having... problems... with servers at work and oom killer [14:59] directhex: err [14:59] it's correct that the OOM killer adjustment is inherited from parent to child [15:00] (most such things are in UNIX) [15:00] I guess for ssh, you don't want bash inheriting it? [15:00] Keybuk, quite [15:01] ArneGoetje: not as far as I can see but surely isn't a big problem since it's done for the delta packages; you can just follow the same approach [15:01] directhex: IIRC there was some later problem as wewll [15:01] well [15:01] sodding linux [15:02] where's my hurd tarball gone [15:02] directhex: and I didn't even know that Novell had taken the *prior* patch I did to support tweaking the OOM killer in sshd at all [15:02] Keybuk: in this context I agree with directhex that it's definitely wrong [15:02] cjwatson: yeah, I've had the same problems with ssh with supervision [15:02] "sshd is still running because someone once ran screen" [15:03] directhex: does Novell set up sshd to run with OOM-killer disabled, then? [15:03] directhex: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=500121 claims to still have the problem, not debugged yet [15:03] Debian bug 500121 in openssh-server "openssh-server: Adjusted OOM killer is inherited by all child processes" [Normal,Open] [15:03] cjwatson, no, they don't [15:03] might be user error but as I say I don't know [15:03] cjwatson, if a user uses lots of ram, then bye bye server [15:03] oh, right, you're running into the *original* problem [15:03] aye [15:04] point them at the upstream bug I guess though I wouldn't blame anyone for not wanting to patch extra features into openssh [15:05] sigh. i just want users who try and use too much ram to have their job killed, and the node ready for the next job. is that so much to ask for? :/ [15:11] i'll poke sgi. if novell aren't interested, sgi can always stick a modified rpm on their repo [15:12]