[03:17] ^ these two (dnsmasq and network-manager) are part of a fix targeted for the point release, so please review as if they were uploaded pre-freeze [03:18] the dnsmasq change should be pretty small, the NM one is likely much bigger. I'll take a look at the diffs myself tomorrow (I've been using the test packages for a few days but didn't look at the code change myself) [07:05] stgraber: you might want to have a look at bug 1033412 [07:05] Launchpad bug 1033412 in lxc "package lxc 0.7.5-3ubuntu60 failed to install/upgrade: unable to install new version of `/usr/lib/lxc/liblxc.so.0.7.5': Device or resource busy" [Undecided,New] https://launchpad.net/bugs/1033412 [07:05] * micahg isn't sure if that's normal or not [07:06] * micahg would think not [08:58] what broke with i386 livefs builds? [10:05] could someone please let my email to ubuntu-release@ through the queue? [10:59] skaet: ^ [11:06] Laney: The machine's down and being recovered. [11:08] righto === yofel_ is now known as yofel [12:54] hi, can anybody reject nvidia-graphics-drivers and nvidia-graphics-drivers from quantal-proposed, please? [12:55] tseliot: You just said the same thing twice. :P [12:55] infinity: err... nvidia-graphics-drivers and nvidia-graphics-drivers-updates [12:55] tseliot: And by reject, you mean remove? [12:56] infinity: if that removes my latest uploads from quantal-proposed then yes please [12:56] Check. [12:57] tseliot: Reason for the removal? [12:58] infinity: I'd like to upload the same revision to quantal and then upload a newer release to -proposed if possible [12:58] That's not gonna happen. [12:58] Once a package version has been accepted, it's "used". [12:58] So, you'll need to upload a new version anyway. [12:59] At which point, it'll supersede this one, so no need to remove it. [12:59] infinity: ok, never mind then. I thought the packages in any -proposed would be pending approval [12:59] Not for the devel release. [12:59] infinity: ok, I've learnt something new, thanks [13:00] It's already in the archive and built. Or, well, failed to build, in this case. :P [13:03] infinity: now that's weird. It doesn't fail in my chroot. I'll investigate the issue... [13:04] tseliot: One failed on one arch, the other on both. [13:06] infinity: right [13:21] micahg: moved to dpkg [13:22] micahg: I have 5-6 servers running lxc that upgraded fine with dozen of containers running on each, so it's definitely not a problem affecting all our users, and I can't see why dpkg wouldn't be able to replace a file that's currently used. [13:27] stgraber: That's not a dpkg bug, can I bounce it back? :P [13:28] stgraber: dpkg is trying to write to the file and failing, it's not up to dpkg to ensure the world is writable. [13:28] stgraber: I'd assume this is either lxc userspace being wonky, or some strange kernel issue, perhaps relating to repeated suspends with active containers? [13:30] infinity: well, "lxc" is only installed outside of the container, so feel free to bounce to kernel ;) [13:31] infinity: I checked the dmesg for anything obvious but without sucess... I guess I could reproduce the failure if the file was a read-only bind-mount, but I don't see why that'd happen [13:31] stgraber: Well, sure, and this is all outside the container. I'm not sure how that's relevant? [13:34] infinity: I don't think this change should be causing a this build failure. Any ideas? https://launchpadlibrarian.net/109173874/buildlog_ubuntu-quantal-amd64.fglrx-installer-updates_2%3A8.960-0ubuntu3_FAILEDTOBUILD.txt.gz http://launchpadlibrarian.net/109008932/fglrx-installer-updates_2%3A8.960-0ubuntu2_2%3A8.960-0ubuntu3.diff.gz [13:36] tseliot: Didn't I already fix that once before? [13:37] tseliot: You need to merge in the changes from fglrx-installer to -updates, I'm thinking. [13:38] infinity: oh, wait, I did that [13:38] infinity: sorry about the noise [14:55] cjwatson, infinity - http://archive.ubuntu.com/ubuntu/dists/quantal/main/dist-upgrader-all/current/ appears to be missing the .html version of the announces. Is there a step that creates *.html automatically, or is manual from nusakan? [14:59] skaet: It would come from the build of the package. [15:05] skaet: Could be that the last upload didn't get pre-build.sh ran before upload or something. [15:07] infinity, yes, looks like prior versions have the .html files, so will assume just temporary glitch and cross check later. [15:07] thanks [15:08] Must be. nusakan can't do this [15:08] * infinity is poking it. [15:08] "It" being the package. [15:09] And any system where this is done manually (though I don't think that existed here) would be very thoroughly deprecated and will soon stop working. [15:12] * skaet nods [15:27] Ugh. release-upgrader's pre-build.sh is just vile. [16:28] It is possible that I should have converted this code into a cleaner language than shell some time ago ... [16:28] Moderately difficult to unpick it into a reasonable object model. [16:58] umm, anything special for seed changes in the point release? see bug 1033575 for context (do we just use the .precise branch as before)? [16:58] Launchpad bug 1033575 in ubuntu-meta "icedtea-plugin shouldn't be shipped on the DVD" [High,In progress] https://launchpad.net/bugs/1033575 [17:00] micahg: yeah, the precise branch is what's used [17:01] ok [17:02] cjwatson: shoop! [17:03] Let's not and say we did. [17:04] I dare you to say you did. [17:04] April 1: Canonical invests millions in shoop development; rewrites all core tools and infrastructure. === Riddelll is now known as Riddell [17:06] Seriously? Another e-d-s ABI bump? [17:07] Maybe we should rename +1 maint to e-d-s-and-poppler-maint. [17:13] I'm going to accept the cups for precise-proposed unless someone objects quickly. It fixes an updates regression. [17:14] ScottK: go ahead, it's been cleared by skaet and I [17:15] OK. Done. [17:17] skaet or stgraber : Assuming we get verification, I any objections to waiving the 7 day waiting period? [17:17] ScottK: let me take a quick look at the fix and I'll answer that one [17:17] OK. [17:19] ScottK: most of the change seems to be about renaming variables and looking at the actual list of affected hardware it looks small enough that even if that one regresses in some way it shouldn't be a disaster, so yeah, I'm fine with that [17:21] if gst0.10-python isn't going to be accepted into precise-proposed due to the freeze then transmageddon should probably be removed (v-failed) [17:21] Laney, would that block the rhythmbox SRU? [17:21] yes [17:22] no way then ;-) [17:24] rhythmbox needs transmageddon needs gst0.10-python [17:25] stgraber: can I get 12.04.1 approval for https://launchpadlibrarian.net/111961124/icedtea-web_1.2-2ubuntu1.2.debdiff ? [17:26] * micahg guesses if he can get Kubuntu to drop it as well, it won't make a difference... [17:27] micahg: sounds like the first fix could wait but the second fixes some upgrade path. It's also not affecting any image we'll be releasing so even if it's slow to validate it won't affect us, so yeah, go ahead [17:28] (IIRC kubuntu won't release a dvd image for the point release) [17:28] it's on the manifest still [17:29] oh, ok, then I guess they actually are planning on releasing a dvd image then :) [17:29] ScottK, Riddell: any opinion on that icedtea-web update? ^ [17:30] * micahg would prefer them to drop it from their DVD as well FWIW :) [17:31] I agree, we should drop it [17:46] I doubt we're respinning the DVD, but I'm checking. [18:09] Actually, we are. [18:09] well I'm happy to be persuaded otherwise [18:10] We didn't before, IIRC. [18:11] I'd suggest respin the DVD once and once only for 12.04.2 after we have 4.8.5 and won't update KDE again. [18:11] That and the usual "Who's going to test it"? [18:12] right, let's do that then [18:12] Riddell: can you update the 12.04.1 manifest then? [18:13] micahg: so it sounds like icedtea-web will no longer affect anything we ship, so no objection to having it accepted [18:14] stgraber: not this one? https://wiki.kubuntu.org/PrecisePangolin/ReleaseManifest [18:14] Riddell: https://wiki.ubuntu.com/PrecisePangolin/ReleaseManifest/12.04.1 [18:15] stgraber: great, will upload later today then, thanks [18:18] skaet: change of plan, no dvds for kubuntu for 12.04.1, manifest updated [18:19] Riddell, ack. We may as well start a 12.04.2 manifest off as well at the same time then, and leave it on there though - right? [18:19] skaet: yep [18:26] stgraber: how do I remove Kubuntu DVD from Precise Daily? I'm looking at http://iso.qa.ubuntu.com/admin/config/services/qatracker/builds and can't work it out [18:26] Riddell: I'll take care of that [18:26] Riddell: starting by disabling the cron on nusakan [18:32] stgraber, while you're on nusakan, can you retrigger Ubuntu Desktop i386 and see if it will build now? Other Ubuntu Desktop images are amd64 at 0806, but that one is at 0804. [18:33] skaet: is the i386 builder back online? [18:33] Riddell: disabled the cron for the dvd image and removed from the tracker [18:34] stgraber, lack of i386 builder would explain it. [18:34] skaet: yeah, I saw a mention of a problem with the i386 builder earlier, explaining all the failures over the weekend. I'm not sure of the state of the fix though. Reading some more backlog now :) [18:35] * skaet starts into the backlog as well. [18:35] skaet: 11:06 < infinity> Laney: The machine's down and being recovered. (7.5 hours ago) [18:37] infinity, is there an RT ticket, or something we can follow for the i386 builder status? === pleia2_ is now known as pleia2 [18:49] Still getting death mails. [19:02] skaet: Not sure, we were just pinged by elmo on the weekend that it was dead and being rebuilt. [19:03] all I see is rt#50669 [19:03] but that implies that it's alive. [19:27] hum [19:27] http://people.canonical.com/~ubuntu-archive/nbs.html is empty [19:27] did we fix everything? ;-) [19:28] heh, possible [19:28] that report has been known to break [19:28] cjwatson fixed it before [19:29] seb128: cross-reference with http://people.canonical.com/~ubuntu-archive/testing/quantal_outdate.html? [19:29] libedata-cal-1.2-15 | 3.4.3-1 | quantal | amd64, armel, armhf, i386, powerpc [19:29] libedata-cal-1.2-18 | 3.5.4-0ubuntu2 | quantal | amd64, armel, armhf, i386, powerpc [19:29] there's an NBS lib right there [19:29] so yeah, the report's broken [19:30] how does one get access to that user to be able to fix it? ^o) [19:32] Laney: lp:~ubuntu-archive/ubuntu-archive-tools/trunk/, cron.NBS [19:32] oh, it's in there [19:33] yep, feel free to fix it without having access to the user account ;) [19:33] (access to the user account should probably be limited to those who are actually archive admins... not sure that you meant to volunteer yourself for that? :) [19:34] Actually I would be interested in that, but perhaps not right at this moment [19:35] seems to be archive-cruft-check that's failing [19:36] http://paste.ubuntu.com/1133105/ [19:42] stgraber: Did you form an opinion about pushing the cups update out as soon as it's verified? [19:46] 17:19 < stgraber> ScottK: most of the change seems to be about renaming variables and looking at the actual list of affected hardware it looks small enough that even if that one [19:46] regresses in some way it shouldn't be a disaster, so yeah, I'm fine with that [19:47] OK. Thanks. I missed that. [19:50] Laney: are you looking into the NBS report breakage? I don't think I'm going to attempt it, given that backtrace [19:50] slangasek: Not exactly. I don't have a local mirror available, so there's some yaks to be shaved first. [19:51] seb128 is an archive admin ;-) [19:51] * seb128 hides [20:04] skaet: so, I'm doing some tests to figure out how we managed to use an extra 8.2MB on the squashfs since 12.04 release [20:04] skaet: there are two obvious ones, the kernel headers (amounting to 2MB) and printer-driver-postscript-hp (around 700KB) [20:05] the rest are packages that just got bigger with the most likely ones being firefox, thunderbird and libreoffice [20:06] stgraber: cups is released. [20:06] ScottK: thanks [20:06] skaet: so, hplip 3.12.2-1ubuntu3.1 is the one that caused the extra 700KB package from printer-driver-postscript-hp (bug 1014478) [20:06] Launchpad bug 1014478 in hplip "PPDs for HP's PostScript printers not installed in default installation" [Medium,Fix released] https://launchpad.net/bugs/1014478 [20:07] so that was clearly done on purpose and approved by the SRU team, so not much hope of getting these 700KB back [20:07] I'll start looking at cutting some langpacks [20:08] stgraber, understood, thanks. Not sure langpacks alone will cut it. Probably time to flag the desktop team then and see if there are good "put package on a diet" suggestions. [20:09] there is none [20:09] have we considered not requiring 12.04.1 to fit on a CD? [20:09] we've already said quantal doesn't have to, and it's coming out in the same time frame; and we don't press CDs of the point releases [20:09] trying to enforce further dieting of the desktop after .0 is fraught with peril [20:10] though, why are the kernel headers so much bigger now? [20:10] slangasek: it's just the live-build bug infinity fixed, we currently ship two header packages [20:11] stgraber: ok; that's been fixed in -proposed since Friday or so, are we not getting test builds yet using it? [20:11] slangasek: nope, we need to poke IS to get it on the builders [20:12] slangasek: infinity ran local tests pre-upload though [20:13] stgraber: ok; will you poke IS please? :) [20:17] slangasek: yeah, infinity said he'd but that didn't happen yet so I guess I can do it ;) [20:19] he's off celebrating August Province Day or something [20:20] skaet: a quick test build here shows that removing the portugese langpack would free us 9MB on the media, making the image fit again. Any strong objection to doing that? [20:20] which I think you're supposed to be as well? :) [20:20] slangasek: nope, we don't seem to have that one on this side of the country ;) [20:20] stgraber, no objection. Much better than alternative. bye-bye portugese langpack [20:21] stgraber: oh yes, infinity did say Quebec was in the exclusion list, ok [20:24] stgraber, skaet: I do object at dropping langpacks this way, how hard did we look for other solutions? [20:25] stgraber, skaet: dropping it makes the iso less useful for 190 millions people, I don't think it's something to overlook [20:26] seb128, weighing this against not having a CD form factor available, it seems the lesser of the evils. [20:26] seb128: well, we need to free 6MB (compressed) from the squashfs, the only package that was added since the release is printer-driver-postscript-hp that according to the bug should really be on the ISO and would only gain us 700KB if removed [20:27] skaet, well, precise was on size, can't we just figure out where we used the space and fix that? [20:27] stgraber: so where did all the rest of the space go? [20:28] slangasek: it's non-trivial to check, my current bet is on libreoffice, thunderbird and firefox, I can try and get a rough estimate of their space before/after (just looking at /usr/lib) [20:28] seb128, that would have been the hope, but it looks like multiple desktop packages all got a bit bigger. [20:28] stgraber: have you looked at .deb sizes, as a first pass? [20:29] pitti said that "/home/cdimage/iso-deb-size-compare on nusakan" can be used on alternate isos [20:29] but I don't have access to nusakan to test that [20:29] that's a different measure than the live image but that can be useful to give a clue [20:31] My assumption is just that firefox and thunderbird grew a ton. [20:31] Not really much mystery there. [20:32] I don't think it's sufficient to assume [20:32] packages putting weight on in SRU are a problem [20:32] binary sizes for firefox and thunderbird grew by around 300KB, will test iso-deb-size-compare see if that helps [20:33] I'm already checking. [20:33] Where do I find the Kubuntu 12.04 dailies? [20:34] ScottK: /kubuntu/precise/ on cdimage [20:34] firefox_11.0+build1-0ubuntu4_i386.deb (17.0 MiB) [20:34] firefox_14.0.1+build1-0ubuntu0.12.04.1_i386.deb (18.0 MiB) [20:34] Thanks. [20:34] http://paste.ubuntu.com/1133208/ [20:35] What's the official size limit for 12.04 CD images? [20:36] ScottK: 736665600 [20:38] * ScottK notices the i386 live is missing. [20:38] The i386 livefs buildd has "issues". [20:38] Apparntly. [20:39] precise released at 732213248 and we are currently at 740601856 (looking for the amd64 desktop image). we should be getting to 737808384 with infinity's change (if my local clean + recompress is right) [20:39] Yeah. I guess "mv: cannot stat `/srv/cdimage.ubuntu.com/scratch/kubuntu/daily-live/tmp/precise-i386/CD1/casper/filesystem.kernel-generic-pae': No such file or directory" is not good. [20:41] so assuming amd64 is the most oversized image, we should just have 1.3MB to cleanup to make it fit again, doesn't look terribly difficult until you remember that's compressed data added post-release we're talking about... [20:42] Kubuntu amd64 is a bit overweight too: 738504704 [20:43] ScottK: this one should fit once the live-build change lands though as it'll save roughly 2 [20:43] MB [20:43] Excellent. [20:46] +printer-driver-postscript-hp 3.12.2-1ubuntu3.1 [20:46] How big is that? [20:46] That's an entirely new package on the current CD. [20:46] 700KB compressed [20:46] I mentioned it earlier, it's bug 1014478 [20:46] Launchpad bug 1014478 in hplip "PPDs for HP's PostScript printers not installed in default installation" [Medium,Fix released] https://launchpad.net/bugs/1014478 [20:54] has somebody fixed the NBS report? if not, I know what the problem is [20:54] cjwatson: I don't believe anyone has - what's the problem? [20:55] basically, bug 459418 [20:55] Launchpad bug 459418 in lazr.restfulclient "Cache is not safe for concurrent use (by processes or threads)" [High,Fix committed] https://launchpad.net/bugs/459418 [20:55] ok [20:55] so occasionally the cache gets corrupted [20:55] so which bits get rm -rfed? [20:55] ~/.launchpadlib/api.launchpad.net/cache - done [20:55] yay [20:56] the second best fix is to make it run with a different cache dir, which I did for sru-report, but it's a bit more fiddly for NBS because there are multiple scripts involved so I've been putting that off [20:56] the best fix is to get somebody in the LP team to release the fix-committed fix for that bug, get it into quantal, and SRU it [21:21] mythtv was granted an MRE and has a 12.04.1 exception, so SRU team members, please feel free to review [21:51] skaet: https://bugs.launchpad.net/launchpad/+bug/174375 comment 27 - do you remember whether the problem was lack of ability to *nominate* to series, or to *target* to series (== approve nominations)? [21:51] Ubuntu bug 174375 in ubuntu-community "Distribution drivers permissions may need redesign" [Medium,Confirmed] [21:52] (the ability to nominate is fairly widespread, and produces a "Nominated for ... [Approve/Decline]" row; the ability to target lets you actually create series-targeted tasks) [21:53] It would make more sense if the problem had been lack of ability to target [21:54] cjwatson, task needed was targetting to a series. [21:54] Right. You'd said nominate in that comment, so I wanted to clarify since they're different permissions in LP. [21:54] yes, bad wording choice on my part. sorry. [21:55] np [22:17] rebuilt Ubuntu with the new live-build, we're now 1796kB short of fitting on the ISO [22:19] stgraber: I'd appreciate it if you'd redo Kubuntu as well so we can confirm if fits now. [22:21] skaet: Would you care to cast your eyes over my latest comment in that bug and make sure I'm not way off base? [22:25] ScottK: sure. I'll trigger amd64 only as i386 is still broken [22:25] ScottK: running [22:26] * cjwatson makes his bimonthly pilgrimage to APUE [22:27] I wonder why cdimage/bin/kill-after bothers to set SIGALRM before forking [22:30] admittedly it does have to set SIGCHLD first [22:40] ScottK: http://cdimage.ubuntu.com/kubuntu/precise/daily-live/20120806.1/ looks happy [22:40] hmm, or rather, it already was happy, might be even happier when the build publishes :) [22:40] (live-build finished but nusakan is still busy building the .iso) [22:55] stgraber: Great. Thanks.