[00:43] cjwatson: robru is internetless for the night, thanks for kicking it off (I'd promised him I would some time around nowish) [01:04] === trainguard: IMAGE 158 DONE (finished: 20140730 01:05) === [01:04] === changelog: http://people.canonical.com/~ogra/touch-image-stats/158.changes === [02:04] === trainguard: IMAGE 159 building (started: 20140730 02:05) === [03:34] === trainguard: IMAGE 159 DONE (finished: 20140730 03:35) === [03:34] === changelog: http://people.canonical.com/~ogra/touch-image-stats/159.changes === [03:36] hello, is anybody here? [03:37] hello bzoltan1, I am. But probably not much use [03:37] bzoltan1: perhaps you can try pinging the train-guards? [03:38] alecu: I wonder if it is possible to get a silo for line38 [03:38] rsalveti: said that he will take care of it during his shift [03:39] bzoltan1: latest qtcreator is already in the archive [03:39] it was a minor patch/bugfix, which was already merged in trunk [03:40] bzoltan1: https://launchpad.net/ubuntu/+source/qtcreator [03:41] rsalveti: ohh man, I am sorry :) You rock! [03:41] rsalveti: yes, it now it was merged :) I was bugging the kubuntu devs about it [04:45] robru, trainguards: I see now that traincon-0 is gone, what's the remaining step for me and silo-10? [07:23] good morning [07:24] sil2100, good morning [07:24] tvoss: morning! [07:35] thostr_: hey! I'm now trying to publish silo 17 and in the packaging diff I see that you guys are modifying existing symbols - will that not break the ABI? [07:37] thostr_: are those symbols used somewhere in external apps? [07:40] sil2100, good morning, so main issue today on 158/159 bug 1350152 [07:40] bug 1350152 in apparmor (Ubuntu) "[mako #158] apparmor denies access to /etc/ssl/openssl.cnf" [Critical,Confirmed] https://launchpad.net/bugs/1350152 [07:41] jibel: hi! I just looked on the smoketesting and saw 2 apps not starting, so this is the reason? It's not a transient error now is it? [07:41] sil2100, no it is not transient [07:41] I saw new uploads for most click core apps in 158, might be related I guess? [07:42] jibel: thanks for bringing this up :) [07:42] sil2100, not sure, because there was no update for gallery-app [07:42] and it's affected by this [07:43] Oh, it is? Since the dashboard is not refreshed for me so I don't see full test results for 159 [07:43] And no gallery-app there [07:46] same problem with camera-app and no update [07:50] Chipaca: hey! [07:50] sil2100: hoi! [07:50] Chipaca: is the merge you want to release reviewed by someone? ;) It's not top-approved (not sure if you have anyone to review it ;p) [07:51] * Chipaca wonders if sil2100 and queuebot are the same person [07:51] sil2100: top-approving.... [07:51] done [07:51] ;p [07:51] Thank you! [07:52] I know selfie top-approvals are not too good, but sometimes there is no other choice..! [07:55] sil2100: in ubuntu-push the branches all go through peeer review before reaching the merge-to-trunk stage [07:55] the only thing i'm approving here is the changelog edits [07:56] Good to know [07:56] sil2100: peer review and automated testing, hence 'automatic'. automatic to trunk involves paperwork and manual testing. [07:57] and if issues are found, back to peer review (usually. if it's a dumb packaging bug i'll fix it and then merge back and have that peer reviewed instead) [07:58] (this happened in this landing, btw; found an issue with one of the dbus endpoints; https://code.launchpad.net/~chipaca/ubuntu-push/fix-clear-persistent/+merge/228712 resulted.) [08:02] sil2100: which reminds me. Are you an ubuntu developer? And if so, could you endorse my PPU application? [08:02] i've been having trouble getting people to go from agreeing to endorse it to actually endorsing it. have had it sitting there for something like two months already. [08:03] Chipaca: I'm a MOTU, not sure if I can but let me take a look at the rules :) [08:03] sil2100: :) ok. [08:04] does anyone know how to get more verbose logs out of ubuntu-app-launch? [08:04] * ogra_ curses ... why do i always forget that [08:05] ogra_, sil2100: somethings come up I should be back for 11, but I'll miss the meeting let me know if you want something dogfooding when I get back [08:06] catch you later guys [08:07] actually here's the syslog from starting calculator: http://paste.ubuntu.com/7902644/ [08:07] apparmor denied [08:07] i guess that's not expected [08:07] seen bug 1350152 ? [08:07] bug 1350152 in apparmor (Ubuntu) "[mako #158] apparmor denies access to /etc/ssl/openssl.cnf" [Critical,Confirmed] https://launchpad.net/bugs/1350152 [08:09] ogra_, somehow i overlooked that when reading victor's email. thanks [08:09] :) [08:10] ogra_, who can fix that one? [08:10] ogra_, also, why no test results for those apps in CI (even if they would be all red) [08:11] hmm, well at least gallery and calculator are not there [08:11] dropping letters and camera are [08:16] brendand, see plars' email [08:16] psivaa, can we get the missing tests onto the dashboard somehow already or are there still issues ? [08:17] ogra_: the uitk went on its infinite loop. so rerunning it again. might take a bit for it to appear. the rest should appear soon. [08:17] ogra_: but with similar blocks of failures as we saw yesterday [08:18] psivaa, thanks ! [08:19] sil2100: @mediascanner: those ABI breaks are not visible outside mediascanner as music app uses qml plugin. But to be 100% sure we're giving it another test round... give us 10 minutes [08:22] bah, the last seed change breaks desktop-next [08:22] thostr_: thanks, just need to make sure about that before I +1 it to the archive, since we don't want ABI breaks without soname bumps [08:24] ah, not the last one but the one before ... hmpf [08:25] ogra_: another breakage in desktop unity8? [08:25] sil2100, well, installing qrtmir-android on desktop unity8 [08:32] hi there, [08:32] I OTA'ed from #133 to #157 but after reboot I'm still on #133 [08:33] I then tried adh shell > system-image-cli -v which also said upgrading to 157 but still 133 after reboot [08:33] /etc/ubuntu-build says 133 too [08:33] known issue ? Any idea ? [08:47] vila, i had no issue [08:48] bzoltan: hi! [08:49] sil2100: hello [08:49] sil2100: what did i screw up again? [08:49] bzoltan: so, we seem to be haunted by some infinite loops from time to time in UITK (rare, but still happening) during dashboard testing [08:49] ogra_: good for you ;-) What can I look/try ? [08:49] bzoltan: oh, nothing! I hope ;p [08:50] bzoltan: but could you guys take a look? psivaa fetched some logs on when the actual loop happens: [08:50] bzoltan: http://pastebin.ubuntu.com/7902557/ [08:50] sil2100: sore [08:50] sure I mean [08:50] vila, hmm, not sure [08:50] bzoltan: this is how the logs looked like, with unity8 being on top from what I remember [08:50] psivaa: ^ is that right? [08:51] vila, there are logs in /cache/recovery/ i think [08:51] sil2100: right, when it happened unity8 was on top [08:54] bzoltan: as always, might not be related to any changes your team did, but you're our first contact here ;) [08:54] sil2100: no worries :) [08:54] ogra_: https://pastebin/canonical.com/114462/ [08:55] bah ... 2fa [08:55] ogra_: hehe, same here ;) [08:55] vila, hmm, looks all fine [08:56] sil2100: That this function used to cause troubles many times. elopio is the one who could tell the real deal [08:56] vila, and "system-image-cli -i" as well as the UI say you are not on 157 ? [08:56] according to that log you should [08:57] ogra_: yes, both say 133 [08:57] sil2100: hm, I'm confused, I thought you were sheriff yesterday but I just checked the calendar and it says that should have been me. Want me to take today? [08:57] vila, weird [08:58] vila, and do you get a new upgrade offered when you click on upgrades in system-settings ? [08:58] ogra_: even if system-image-cli says: last update: 2014-07-30 10:55: 29 (I retried an OTA) [08:58] ogra_: yes, for 157 [08:58] vila, and did you try to run it again ? [08:59] cjwatson: ok, sure thing, would be grateful as I have some paperwork to finish and then doing some real test landing and tweaks for the other-distro bits of citrain [08:59] ogra_: as of now, 2 OTA, 2 system-image-cli === cjwatson changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard (general help): cihelp | CI Train support: cjwatson | CI Train Status: #157 promoted | Known issues: Read http://bit.ly/1k4uBJR before asking about the spreadsheet changes. Both queuebot and http://bit.ly/1mDv1FS know your silo status before the spreadsheet does. [09:00] sil2100: ok, anything I should know from the meeting? I guess we can mostly publish fairly freely ... [09:01] ogra_: trying 'system-image-cli -b 0 -v' just in case [09:01] cjwatson: yeah, normal landings resumed - we're concentrating on getting the current new blockers resolved (the apparmor issue making application launch problems) but other then that, it's normal [09:02] cjwatson: we're thinking of kicking a new image soon anyway as there has been a change to lxc container handling that we'd need tested [09:02] ogra_: got channel ubuntu-touch/utopic, upgrade path is 157 [09:03] vila, sounds right [09:03] perhaps the file is corrupt or some such ... you coudl try to wipe /cache/recovery/ [09:03] that should force a re-download [09:03] ogra_: when is /etc/ubuntu-build ? [09:04] pete-woods, ping [09:04] thtas put in place as last step during the upgrade iirc [09:04] ogra_: the file is not there after reboot, only the logs (will triple check once that -b 0 finish) [09:06] pete-woods, ping [09:11] pete-woods,tvoss: any reason I can't publish silo 8? I see it conflicts with silo 1 [09:12] cjwatson, not from my side, pinged pete-woods but he did not come back to me, yet [09:12] ogra_: doh, it worked ! Thanks for the debugging vibes or whatever trick you used ;) May be it's just because it rebooted at 11h11... [09:13] haha [09:14] sil2100: can you reconfigure silo 17 [09:14] thostr_: I'll do it [09:14] thostr_: (see topic) [09:15] ack [09:15] thostr_: conflicts with silo 5 [09:15] tvoss: hi [09:15] thostr_: can you agree this with Saviq/mhr3? [09:16] pete-woods, are we good to go for silo 8? [09:16] given that silo 5 builds right now let's land that first then [09:16] thostr_, go for it, we'll manage [09:16] thostr_, 5 won't land until tomorrow at least [09:16] Saviq: ok, then I'll take the slo [09:16] tvoss: yep [09:16] cjwatson: ok, then silo 17 first [09:16] sooner the better really, it fixes a serious scope bug [09:18] cjwatson, ^ [09:18] cjwatson, good to go for silo 8 [09:18] cjwatson, will rebuild silo 1 after that [09:18] Saviq: the /wizard/, maybe - but u-s-s itself ought to work [09:21] thostr_,tvoss: ok [09:25] Saviq: do you know why this dependency isn't on -wizard, but on u-s-s itself? [09:47] mterry: ^-- could you check? [09:47] Morning all [09:47] sil2100: you want me to dogfood something? [09:48] davmor2: hey! Not yet ;) [09:48] Laney, read scrollback to see that some dep of USS should be on the wiz, but which one? [09:48] davmor2: the current image has some issues that we need jdstrand for [09:48] Laney, wiz used to be in same binary package, so I wouldn't be surprised [09:48] sil2100: oh that doesn't sound good [09:49] mterry: the qtmir thingy [09:50] sec [09:50] sil2100: Are we waiting for the security folks to wake up before addressing this apparmor bug? [09:50] Laney, yeah... that could be on the wizard only I believe [09:51] cjwatson: we know that jdstrand was working on it yesterday, and so far he's our only contact regarding these issues [09:51] mterry: It might be sensible to do ... (one second) [09:51] mkay [09:52] https://bugs.launchpad.net/dropping-letters/+bug/1350152 <- this is the bug (or at least one manifestation of it) [09:52] Launchpad bug 1350152 in apparmor (Ubuntu) "[mako #158] apparmor denies access to /etc/ssl/openssl.cnf" [Critical,Confirmed] [10:03] mterry: https://bazaar.launchpad.net/~laney/ubuntu-system-settings/qtmir-wizard/revision/844  [10:04] Laney, it used to theoretically support non-mir mode by dynamically loading unity-mir. I see that with the change to qtcomp, there is still a check for QT_QPA_PLATFORM and a variable called isMirServer, which makes me think we still support non-mir architectures. greyback? [10:05] not just theoretically -- I ran it in non-mir [10:05] why was this added then? [10:05] this happened just yesterday [10:05] Laney, where supported, we want to use Mir [10:06] Laney, though I suppose that c(sh)ould be a recommend rather than depend [10:06] the first attempt was a build-dep to force it to not build on non-qtmir arches [10:07] cjwatson, does my last touch seed change need more than a publisher run ? [10:07] * ogra_ doesnt think so but i want to be sure [10:11] Laney, greyback is about to sit down, he can explain whether we need it as a recommend or depend anyway [10:11] Laney, I'm not sure what the problem we're causing with the dep is right now? [10:12] mterry: It's making my allergy flare up [10:12] Laney, greyback confirmed in verbal talks that non-mir fallback still works [10:12] Laney, so we can leave Arch as any [10:12] ogra_: We don't use tasks at present for touch, so it doesn't need a special publisher run but does need a metapackage update [10:12] mterry: so Recommends? [10:12] assuming something else pulls it in [10:13] or the arch restricted dep if not [10:13] cjwatson, ah, heh, i thought it was exactly the other way round ... will do a meta rebuild then [10:13] ogra_: Oh, but desktop-next uses tasks [10:13] Laney, sure I'm fine a Recommends, seems cleaner [10:13] ah, good [10:13] ogra_: You may be right, I guess this has no effect on the metapackages ... [10:13] ogra_: So yeah, just give it a couple of publisher runs to settle [10:13] good [10:14] mterry: yay, thanks, MP in a minute [10:14] Laney, sorry to irritate your allergies :) [10:14] just want to have desktop-next buildable again [10:14] mterry: wasn't you - I traced it to https://code.launchpad.net/~unity-team/ubuntu-system-settings/use-qtComp/+merge/225540 [10:14] Laney, yeah but I think I reviewed the packaging to that [10:14] I bet it made u-s-s uninstallable and so angered proposed-migration to which the rest of the MPs were a response [10:14] Laney, and didn't notice the arch problem [10:15] ah well [10:15] Laney: correct [10:15] hilarious, everyone at the time was insisting that it was totally broken without mir [10:16] Definitely not the main application or else I don't know what I'm developing on. :) [10:17] I would have believed someone saying that about the wizard though, but this is in a different binary package anyway [10:20] Chipaca: ok, so I checked and most probably normally I should be able to endorse your application, but hm, it doesn't seem I sponsored any packages for you! [10:21] Chipaca: most probably because when I was doing that, it wasn't noted as a sponsor by me [10:21] sil2100: isn't every landing a sponsorage? [10:22] Chipaca: now it is, before we weren't mentioned as sponsors of a package... so for now I guess Ubuntu will notice only one sponsoring made by me [10:22] And besides, I am a MOTU since over a month only, so not too long [10:22] sil2100: wfm :) [10:22] Chipaca: in any case, could you send me a link to your application? ;) [10:22] sil2100: maybe. https://wiki.ubuntu.com/Chipaca/PPU [10:34] Laney, did you mean to leave the "Architecture: any" removal in your merge? [10:35] mterry: nope [10:35] look now [10:37] sil2100, any update on the propagation status of silo 8? [10:42] tvoss: please ask the current sheriff rather than bothering sil :) [10:43] cjwatson, hey there :) any update on the propagation status of silo 8? [10:43] tvoss: unity-scopes-{api,shell} are waiting for autopkgtests to run [10:43] No worries, not bothered ;p I'm just busy with a package right now, so sorry for not being responsive! [10:43] cjwatson, ah :) [10:44] tvoss: there's something of a queue (http://d-jenkins.ubuntu-ci:8080/view/Utopic/view/AutoPkgTest/) so might take a bit, but nothing appears to have gone wrong as yet [10:45] cjwatson, cool, thank you [10:47] cjwatson: can I get a silo for line 31 [10:48] thostr_: as it happens I was just in the process of assigning that [10:48] :) === MacSlow is now known as MacSlow|lunch === alan_g is now known as alan_g|lunch === tvoss is now known as tvoss|lunch [11:46] hi [11:46] so, bug #1350152 is a new bug [11:46] bug 1350152 in apparmor (Ubuntu) "[mako #158] apparmor denies access to /etc/ssl/openssl.cnf" [Critical,Confirmed] https://launchpad.net/bugs/1350152 [11:46] it was not seen with 157 and is entirely different than the two issues seen yesterday [11:47] * jdstrand notes that there were 3 entirely different apparmor denials [11:47] jdstrand, did anyone claim it was ? [11:47] yes [11:47] oh [11:47] (surely isnt) [11:47] in another email thread and it was mentioned in this channel that I was working on this [11:48] heh, how would you ... the issues were only found while you slept :) [11:48] I didn't see this bug yesterday and only just saw it a moment ago [11:48] exactly :) [11:48] anyway [11:48] something in the platform changed [11:48] it seems for all apps ssl is involved but they seem to look in different locations [11:48] we have the openssl abstraction in the networking policy group [11:49] apps that don't have the networking policy group are seeing this denial [11:49] ie, they are using openssl now when they didn't used to [11:49] http://people.canonical.com/~ogra/touch-image-stats/158.changes [11:49] gnutls update ? [11:49] whole batch of transitions landed [11:50] you wouldn't expect gnutls to change openssl things though! [11:50] well, yeah [11:51] (and no matches for openssl.cnf in its diff) [11:51] it might be easier to track down directly with strace or similar [11:51] I don't have an up to date emulator handy [11:57] * mterry wants a new image [11:58] sil2100, where do we stand with that ? ^^ [11:58] (i assume the apparmor triaging will still take a while anyway) [11:58] * jdstrand starts regenerating an emulator image, but it is going to take hours [11:58] yeah [11:58] Ok, if that's the case, let's build an image now [11:58] where do we stand on system-image updates with the emulator? [11:58] We can build one before the evening meeting [11:59] (in case some fixes fly in) [11:59] ogra_: should I kick the image? [11:59] sil2100, go for it :) [11:59] jdstrand, probably an rsalveti question [11:59] jdstrand: just to make sure.. i dint claim it was the same case, but asking if it was likely.. apologies if that caused further cofusion [11:59] * ogra_ has to much HW to use the emulator :P === psivaa is now known as psivaa-lunch [11:59] there was no confusion on my end [11:59] :) [11:59] it isn't the same issue [12:00] sil2100, I want one to test a lxc-android-config change before I can push in locking support [12:00] psivaa-lunch: no worries [12:00] right not sure if there was confusion about my question though [12:00] mterry: right :) We also wanted to have an image with the new lxc-android-config, but wanted to know if the apparmor issue is, by any chance, not a 'quick 5 minute fix' [12:00] sil2100, gotcha [12:01] mterry: an image should be building now [12:01] * mterry pumps fists === cprov changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard (general help): cprov | CI Train support: cjwatson | CI Train Status: #157 promoted | Known issues: Read http://bit.ly/1k4uBJR before asking about the spreadsheet changes. Both queuebot and http://bit.ly/1mDv1FS know your silo status before the spreadsheet does. [12:02] mterry, i assume with that we can finally land the locking stuff ? [12:02] ogra_, right, assuming everything still checks out [12:02] it will, it will :) [12:02] I wonders if it is curl [12:02] heh, I'm much more pessimistic than you, after my last few months of trying to land things :) [12:02] * ogra_ has a hammer in the drawer in case it wont :) [12:04] === trainguard: IMAGE 160 building (started: 20140730 12:05) === [12:05] such a nice rround number [12:05] ogra_: bound to go wrong then ;) [12:05] davmor2, well, we knoe some apps wont start :) [12:06] Apps are sooo passe [12:06] heh [12:07] davmor2, this is for testing that the system still works after upgrading ... we moved /etc/passwd|group|shadow arouns a last time ... [12:07] *around [12:07] (teh proper fix for the quick hack i added initially when we couldnt boot anymore) === MacSlow|lunch is now known as MacSlow [12:09] ogra_, sil2100: so what happened between 157 and 158 to break the universe, /me prepares to blame that ogra_ bloke if it isn't him it'll be that popey bloke [12:09] davmor2, read mails :P [12:09] not me guv [12:09] plars wrote one ;) [12:09] ogra_: haven't got that far through my mail stack yet :D [12:10] sjust start from the top, not the bottom :P [12:11] tvoss|lunch: silo 8 merged, so you can resync 1 now [12:12] davmor2: don't worry, I'll go back and get results for 158 if needed. since 160 is about to hit us, let me try to get fixed credentials for the dashboard first if that's ok :) [12:12] plars: no rush dude I'm gonna get to your email [12:12] plars, i think we can skip 158 ... wont be much different from 159 [12:13] (there was only a seed change between the two) [12:13] ogra_: when I got it up and running, I think 159 had just hit and I didn't see that I was at a transition there, or I would have started both before going to sleep [12:14] no worries [12:37] Hm, I see that the RTM branch procedure didn't include ofono-phonesim, but that's needed for some autopilot tests [12:38] Is there a seed anywhere that includes that stuff? [12:38] Also uoa-integration-tests [12:39] Otherwise I might create one just to help with the branching === psivaa-lunch is now known as psivaa === Ursinha changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard (general help): Ursinha | CI Train support: cjwatson | CI Train Status: #157 promoted | Known issues: Read http://bit.ly/1k4uBJR before asking about the spreadsheet changes. Both queuebot and http://bit.ly/1mDv1FS know your silo status before the spreadsheet does. [13:11] where can I get the click for the calculator? === alan_g|lunch is now known as alan_g [13:12] jdstrand: [13:12] http://people.canonical.com/~ubuntu-archive/click_packages/ [13:13] oh, that is convenient [13:13] thanks! [13:21] ogra_: could you check out https://code.launchpad.net/~cjwatson/ubuntu-seeds/touch-tests/+merge/228846, please? [13:22] I've run germinate over it and it seems to be producing basically plausible results [13:22] http://paste.ubuntu.com/7904805/ <- the touch-tests result file === tvoss|lunch is now known as tvoss [13:25] cjwatson, approved [13:26] star, thanks [13:27] pete-woods: https://code.launchpad.net/~pete-woods/net-cpp/uri-builder/+merge/222614 looks like the changelog needs to be remerged/reordered [13:27] doesn't look right for that change to be inserted into the middle [13:28] cjwatson, is there a way to install a file via packagename.install only for specific architectures? [13:28] several [13:29] you can preprocess the file, or sometimes dh-exec works, or you can use foo.install. files [13:29] do you have an example? [13:29] https://launchpad.net/~ci-train-ppa-service/+archive/ubuntu/landing-001/+packages [13:29] === trainguard: IMAGE 160 DONE (finished: 20140730 13:30) === [13:29] === changelog: http://people.canonical.com/~ogra/touch-image-stats/160.changes === [13:29] fails on archs that don't have android gps HAL support as the respective provider does not get compiled in, and not even the config file /etc/gps.conf is installed [13:30] cjwatson, however, /etc/gps.conf is in debian/libubuntu-location-service2.install [13:30] didnt dh_install have an option to use the arch name for the .install file ? [13:30] seems the manpage doesnt know about it [13:30] * mterry installs image 160 [13:30] but i seem to remember having used it before [13:31] ogra_: that was one of my suggestions above. please leave this to me? [13:31] yeah [13:31] i was just wondering why the manpage is silent about it [13:31] it's awkward to have two people giving advice :) [13:31] ogra_, sil2100: so 160, what should it fix and is it worth my time looking at it? [13:31] it's in debhelper(7) because it's a general facility, not specific to dh_install [13:31] oh, right ! [13:31] thanks [13:32] davmor2, it should only prove that you still have a running system after upgrade (like i said above ... we are testing the passwd|shadow|group change with it) [13:33] ogra_: right okay so I ota and should have a system that boots right? [13:33] tvoss: ok, give me a minute [13:33] davmor2, right [13:33] cjwatson, thank you :) [13:33] davmor2, and the same on a fresh flash indeed [13:36] tvoss: ok, so there are two sensible alternatives here, and I'll leave it up to your taste which you choose [13:36] ogra_: Good news I am into unity8 on mako [13:36] ota [13:37] tvoss: you can revert the change to debian/libubuntu-location-service2.install, create debian/libubuntu-location-service2.install.with-gps with the extra entry, and symlink debian/libubuntu-location-service2.install.amd64, debian/libubuntu-location-service2.install.armhf, and debian/libubuntu-location-service2.install.i386 to that [13:37] kenvandine, heyo, if I wanted to land my u-s-s locking branch today, any conflicts? [13:38] tvoss: or you can move debian/libubuntu-location-service2.install to debian/libubuntu-location-service2.install.in and arrange to preprocess it with sed or similar, by adding a make rule to debian/rules to generate it and having override_dh_install depend on that new target; you'd also want to add debian/libubuntu-location-service2.install to debian/clean in that case [13:38] cjwatson: updated :) [13:38] ohhhhh I like the icon for the app in the start up frame with the spinner :) [13:39] tvoss: (silo 8 ready for you, if you didn't notice) === Ursinha is now known as Ursinha-afk === pete-woods is now known as pete-woods|lunch [13:40] cjwatson, thank you, I will go for option one. and thanks for the silo, will finish off silo 1, first, though [13:40] tvoss: ok [13:41] davmor2, awesome [13:41] mterry, are you ready now? [13:41] pete-woods|lunch: surprised to see no symbol file updates in that branch. are you planning to just get that from the first build failure or something? [13:41] mterry, i'll have some stuff i want to try to land later today [13:41] kenvandine, locking ? [13:42] mterry, but not all ready yet, so go ahead [13:43] kenvandine, I'm basically ready. Final testing [13:43] mterry, sweet [13:43] ogra_, ask mterry, i'm just know it has to land along with some other stuff [13:43] cjwatson: at least when I made that branch, net-cpp didn't do symbol tracking [13:43] perhaps that's changed, though [13:44] * pete-woods|lunch re-running bzr bd [13:44] pete-woods|lunch: oh, seriously? ugh, ok [13:44] it's not my project :p === pete-woods|lunch is now known as pete-woods [13:46] okay, it has changed now, will update the symbols file [13:50] sil2100, messaging-app is ok, but there is a definite issue with notes-app. i need to file a bug [13:52] cjwatson: okay, have added the symbols now === pete-woods is now known as pete-woods|lunch [13:53] brendand, you mean beyond the apparmor denial ? [13:53] ogra_, notes didn't have an issue with that [13:54] ah, k [13:54] ogra_, at least it doesn't crash on start [13:56] robru, heyo! I think silo-012 is ready to land [13:57] sil2100, ^ [13:57] cjwatson, ^ I haven't started my shift yet ;-) [13:57] pete-woods|lunch: thanks. sorry, yet another thing - wouldn't it be usual to bump the minor number for new interfaces rather than the patch number? That would make this 1.1.0 rather than 1.0.1 (more importantly, NET_CPP_VERSION_MINOR 1 and NET_CPP_VERSION_PATCH 0) [13:58] pete-woods|lunch: (I didn't mention that before since wasn't certain that there was new ABI involved, but there certainly is if you have to change .symbols) [13:59] robru, sorry man :) [13:59] robru, I'm in London, my timing is all messed up === Ursinha-afk is now known as Ursinha [13:59] mterry, he is in europe too ... just pretending to not work :P [13:59] robru, oh there's a CI train support person in the title now [13:59] mterry, I'm in Strasbourg, my timing is messed up too ;-) But I know my shift doesn't start for 2 more hours ;-) [13:59] mterry, yeah [14:00] well maybe sil and cj have plans for what to land, I don't want to step on their toes until after the daily hangout === plars changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard (general help): plars | CI Train support: cjwatson | CI Train Status: #157 promoted | Known issues: Read http://bit.ly/1k4uBJR before asking about the spreadsheet changes. Both queuebot and http://bit.ly/1mDv1FS know your silo status before the spreadsheet does. [14:04] brendand: thanks! [14:10] cjwatson, could use some help here: https://ci-train.ubuntu.com/job/landing-001-1-build/152/console [14:10] tvoss: symlinks in debian/ only work in 3.0 source formats [14:10] cjwatson, oh ... adjusting [14:11] tvoss: can you make this be 3.0 (quilt)? [14:11] sorry, forgot to mention that, everything I maintain has been 3.0 for years :) [14:11] cjwatson, I only need to adjust source/format, correct? [14:13] tvoss: I believe so; there's no manual patch system junk to undo in this case, or anything. I'm less familiar with whether bzr bd will be happy, but hopefully it should be [14:13] (have mostly stopped using bzr for packaging ...) [14:13] cjwatson, ack, just giving it a spin locally [14:23] cjwatson, I forget what's involved in landing a silo, but I'd like to initiate the process for silo 012 [14:23] presumably it's as simple as pressing the publish button? [14:26] mterry: yes, though I'd have done it for you if I hadn't been distracted - I'll just do it now [14:26] cjwatson, oh OK thanks [14:26] * mterry is excited to see this land finally [14:28] lock 'n loll !!! [14:31] ogra_, :) [14:31] kgunn, ^ [14:36] anybody know what's up with this build failure? https://launchpad.net/~canonical-ci-engineering/+archive/ubuntu/ci-airline-phase-0/+build/6222466 is this an archive problem and do we need a no-change rebuild of zope.proxy? [14:38] barry: does it fail against plain trusty locally? I notice that that PPA contains zope.proxy [14:39] barry: so I'd guess it to be a problem local to that PPA [14:39] cjwatson: yeah, it's probably local to the PPA because z.untrustedpython isn't in trusty [14:40] cjwatson: do we know when the official stable-proposed will happen? I don't think there have been any updates to the staging one right? [14:40] i guess i won't worry about it until someone!=me complains :) [14:40] plars: see thread on ubuntu-phone@ [14:40] ogra_, sil2100, davmor2: we have live results again [14:41] yay [14:41] plars: \o/ [14:41] Thanks :) [14:41] plars: I was about to copy a few more packages into the current RTM branch and ask for a rerun of a few tests in touch_stable, but wasn't planning to do an image rebuild [14:41] ok, finally have a emulator up for the openssl.cnf denial [14:42] cjwatson: ok [14:43] oh, why is touch_stable so much less broken ? [14:43] must be the name [14:44] ogra_: it was branched from some time ago [14:44] ah [14:45] there are three test suite failures that are down to the touch-tests branch. The other tests, no idea, they don't look like they're related to the archive branch so I'm not planning to investigate much [14:55] I just pushed the 160 and many apps do not start... calculator, camera [14:57] bzoltan1, read your emails :P [14:57] (we have apparmor issues for a handfull of apps) [14:59] ogra_: I missed that mail [14:59] bzoltan1, jdstrand is looking into it now [14:59] pmcgowan: thanks [15:00] bzoltan1, apps are accessing openssl.cnf for some reason [15:00] ping trainguards. Can I ask for a silo for click-scope, row 34? [15:00] bzoltan1, anything added to the toolkit to do that? [15:00] alecu: I haven't assigned that because it clashes with silo 5 [15:02] pmcgowan: not as I know [15:05] cjwatson: thanks for the heads up, I'll see about solving that. [15:05] alecu: if the owners of that silo will be landing after you and are happy to rebase, I'll assign, but please check first [15:06] sure [15:11] cjwatson: will ping tvoss for his preference :) [15:12] pete-woods|lunch: thanks, sorry for the back-and-forth === pete-woods|lunch is now known as pete-woods [15:13] cjwatson, it's fine to give the override to the apps scope, already done that for the mediascanner one [15:15] ok, if I downgrade gnutls and curl, there is no denial [15:15] 3.2.15-1 and 7.36.0-2ubuntu1 [15:16] cjwatson: okay, that's updated now, too [15:16] hopefully I will not make further trivial errors :) [15:18] jdstrand, interesting, whats the tie in to an app? [15:19] jdstrand: can you see exactly where in that chain? I don't see anything in either diff [15:20] sil2100: ooh [15:22] * sil2100 is in the middle of eliminating all the small typos he's been making just now [15:22] cute ! [15:23] pmcgowan: I don't know yet [15:23] rsalveti, ping [15:23] cjwatson: I haven't been able to yet [15:23] Neither gnutls28 nor curl mentions openssl.cnf in its source [15:23] I guess something is calling out to openssl for some reason? [15:24] rsalveti, just checked syslog, no more denies on /etc/gps.conf with silo 1, setting tested to true [15:29] cjwatson, jdstrand i see there are fixes for the nss build in curl ... could it be related to i.e. calling getpwent() by the app which then goes through nss with somehow then (for whatever reason) looks for ssl stuff ? [15:29] (as you might see i'm wildly guessing here) [15:29] no idea [15:29] cjwatson: I did strings on all the libs, the only thing that references it is /lib/i386-linux-gnu/libcrypto.so.1.0.0 [15:29] this is why I suggested strace [15:30] I tried strace, but I lack the fu to see what library is doing the access [15:30] jdstrand, is that with every app? [15:30] tvoss, only a bunch [15:30] jdstrand: ltrace -S might be worth trying too, if it works [15:30] gallery, camera ... [15:30] (IME ltrace is not the most reliable tool in the box, but useful when it does work) [15:31] tvoss: as it happens, the networking policy group specifies the openssl apparmor abstraction, which allows that access. in other words, this only affects apps that do not specify the networking policy group [15:31] jdstrand, hmmm ... which apps do try the access? [15:31] jdstrand, just straced the camera app, did not see the open call [15:31] they are listed in the bug. calculator is the one I am looking at [15:32] I'm told camera-app too [15:32] and gallery [15:32] tvoss: you need at least image 158 on arm. I'm using 162 on x86 [15:33] jdstrand, I'm on the latest image here [15:33] tvoss, and gallery starts for you ? [15:33] https://bugs.launchpad.net/gallery-app/+bug/1350152 [15:33] Launchpad bug 1350152 in apparmor-easyprof-ubuntu (Ubuntu) "[mako #158] apparmor denies access to /etc/ssl/openssl.cnf" [Critical,In progress] [15:33] ogra_, nope, just tried [15:34] right [15:34] ogra_, is the denial the cause of the startup issue? [15:34] tvoss: latest promoted or latest proposed? latest promoted doesn't have the issue [15:34] tvoss, seemingly [15:34] tvoss, you coudl disable apparmor and test for laughs ... [15:35] ogra_, interestingly, starting the app from the command line works [15:36] tvoss, well, i guess that doesnt go through UAL [15:37] jdstrand, cjwatson I don't get a denied in /var/log/syslog when starting gallery from the command line [15:37] because you dont call aa-exec which UAL does [15:37] (i guess) [15:38] ogra_, ah yeah ... true === Ursinha is now known as Ursinha-afk [15:39] strace and frineds should still show you if/when it tries to open the file [15:39] tvoss: how are you starting it? [15:40] * jdstrand is using: /tmp/trace aa-exec-click -p com.ubuntu.calculator_calculator_1.3.291 -- qmlscene -qt5 ubuntu-calculator-app.qml --desktop_file_hint=~/.local/share/applications/com.ubuntu.calculator_calculator_1.3.291.desktop [15:40] (from within phablet-shell) [15:40] err [15:40] get rid of that first /tmp/trace :) [15:41] aa-exec-click -p com.ubuntu.calculator_calculator_1.3.291 -- qmlscene -qt5 ubuntu-calculator-app.qml --desktop_file_hint=~/.local/share/applications/com.ubuntu.calculator_calculator_1.3.291.desktop [15:41] (that is after a cd to /usr/share/click/preinstall/com.ubuntu.calculator/current [15:41] jdstrand, phablet@ubuntu-phablet:/usr/share/click/preinstalled/com.ubuntu.gallery/current$ strace ./gallery-app --desktop_file_hint=/usr/share/applications/webbrowser-app.desktop 2> /tmp/strace.txt [15:42] jdstrand, interestingly, that gives me a call to open on open("/usr/lib/ssl/openssl.cnf", O_RDONLY|O_LARGEFILE) [15:42] tvoss: that doesn't run under confinement [15:42] jdstrand, right [15:43] jdstrand, still interesting that the access is different with no confinement [15:46] fginther, Hi! Can you please tell whats going wrong here: https://jenkins.qa.ubuntu.com/job/autopilot-testrunner-otto-utopic/1802/console ? [15:46] I see its extracting address-book-service-testability deb but when it comes to installing it its not able to find the package [15:48] om26er, does this happen to be a new package that isn't in the archive yet? === Ursinha-afk is now known as Ursinha [15:48] fginther, yes [15:51] om26er, I was wondering when this was going to bite us. The test tries to install packages from the archive first before looking at the local packages. I might be able to find a fix [15:52] ltrace showed nothing [15:52] jdstrand, \o/ [15:53] I'm inclined to just allow the access in the templates since allowing the access allows the app to start, but it would be nice to know why this happened [15:53] (and the app didn't magically need networking) [15:53] jdstrand: ogra_: xnox was working on emulator updates (system-image), not sure what is the current status though [15:55] * jdstrand nods [15:58] ogra_: hm, I need to jump out somewhere right now, could you lead the meeting? [15:58] so, if I downgrade just curl, I don't see it, but that doesn't mean it isn't in gnutls since the new curl needs the new gnutls [15:58] ogra_: just give me a sign on the updates [15:59] Curl_ossl_init: call OPENSSL_config for initing engines [16:00] jdstrand: sounds scary. If gallery app tries to open openssl.cnf, it's probably got transitively linked to openssl for which most of our apps to do not have exception. (all but notes-app) [16:00] I think it is this: http://curl.haxx.se/mail/lib-2014-06/0003.html [16:00] xnox: I think this has to do with the compat stuff [16:00] * sil2100 jumps out now [16:01] jdstrand: should die with fire, as no need on ubuntu-touch? one can't and shouldn't modify non-default openssl stuff =) [16:01] jdstrand: or like, handle the denial / initialisation gracefully. [16:01] xnox: no on ewants to modify, it is just a 'r' === gatox is now known as gatox_lunch [16:04] cjwatson, I'm confused, what am I doing wrong here? https://ci-train.ubuntu.com/job/landing-008-1-build/185/console [16:07] tvoss: that's due to https://launchpad.net/ubuntu/+source/dbus-cpp/4.0.0+14.10.20140718-0ubuntu2 - you could merge that manually onto lp:dbus-cpp first (just a changelog delta), or since it's just a no-change rebuild we could ignore that. which would you prefer? [16:08] cjwatson, how do I merge manually? [16:09] tvoss: oh, I can do it [16:09] just a direct push to lp:dbus-cpp, but since I apparently have access I'll sort it out [16:10] cjwatson, would be great to get trunk in sync with distro :) [16:11] tvoss: done, you can retry [16:12] cjwatson, looking good [16:13] cjwatson, for silo 1, any eta on the migration? [16:13] sil2100, ogra_ - notes-app bug https://bugs.launchpad.net/notes-app/+bug/1350361 [16:13] Launchpad bug 1350361 in notes-app "test_note_focus_on_second_click_outside failed on mako #159" [Undecided,New] [16:13] sil2100, ^^^ for the mail [16:15] tvoss: the publisher seems to be particularly slow at the moment - I'll look into that [16:15] tvoss: it's still publishing to -proposed [16:15] * tvoss hugs the poor publisher [16:16] I've updated the bug and am working on an update for apparmor-easyprof-ubuntu now [16:17] tvoss: I've seen this kind of thing before - I think it just needs an ANALYZE. have asked webops for that [16:18] * ogra_ hugs jdstrand [16:18] awesome ! [16:25] jdstrand: already man you rock \m/ [16:25] thanks! [16:26] ok, publisher hopefully faster now [16:26] I'll upload it after I'm done running the test plan [16:26] robru, perhaps wait with an image build for that fix [16:26] ogra_, alright [16:26] pointless ~10-minute delays FTL [16:26] ogra_, so I just tried opening some apps. only the most recently used app live updates in the switcher. [16:26] jdstrand: does the testplan now include opening the apps that didn't open? [16:27] robru, right, only the running one [16:27] ogra_, i thought you said they all did? that would be *way* cooler ;-) [16:27] jdstrand: daft question I understand but it might be worth adding for future test runs [16:28] davmor2: no. but it does include running the smoke image tests, which test for various file access, which I just updated for this [16:28] robru, heh, it would look good but eat your battery :) [16:28] davmor2: I of course did test that the updated package fixes it for the calculator [16:28] in other words-- yes, I am testing it and adding a test for the future [16:28] jdstrand: thats cool then as long as it was covered for future releases :) [16:29] it is [16:29] oh, we have -dev3 now [16:31] jdstrand, do you need a silo for that or are you just going to upload it directly? [16:36] robru: will upload directly [16:36] jdstrand, cool, gimme a ping when that's done so I can kick an image. thanks! [16:37] sure thing [16:40] fginther, I've now branched mir/0.6. would you mind setting it up for jenkins? === robru changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard (general help): plars | CI Train support: robru, trainguards | CI Train Status: #157 promoted | Known issues: Read http://bit.ly/1k4uBJR before asking about the spreadsheet changes. Both queuebot and http://bit.ly/1mDv1FS know your silo status before the spreadsheet does. === robru changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard (general help): plars | CI Train support: robru, trainguards | CI Train Status: #157 promoted | Known issues: Both queuebot and http://bit.ly/1mDv1FS know your silo status before the spreadsheet does. [16:41] robru: thanks [16:41] cjwatson, you're welcome! [16:50] gah, publisher still slothful, am out of ideas. if it's still like this next week then I'll see if I can work with William to improve things [16:50] it might need the new DB servers plugged in before we can improve it much :-/ === alan_g is now known as alan_g|EOD === gatox_lunch is now known as gatox [17:07] robru: apparmor-easyprof-ubuntu 1.2.13 uploaded [17:07] jdstrand, thanks! [17:07] np [17:10] plars: Could you please re-run the online_accounts_ui, dialer_app, and messaging_app suites in touch_stable (for all platforms that have each one)? I copied some more packages into the dry-run archive, so those suites should be happier now. [17:14] cjwatson: will do [17:29] jdstrand, was your upload accepted? I don't see it in proposed. [17:32] tvoss, alecu, Saviq, kenvandine: anybody around to use a silo if I give you one? I have a couple. [17:32] boiko, ^ [17:33] robru: me! :) [17:33] boiko, ok ;-) [17:33] robru, soon :) [17:36] :D [17:38] robru: it is listed here: https://launchpad.net/ubuntu/+source/apparmor-easyprof-ubuntu/1.2.13 [17:38] it built [17:38] https://launchpad.net/ubuntu/+source/apparmor-easyprof-ubuntu/+publishinghistory [17:39] shows it pending for proposed [17:39] I think I saw somewhere that the publisher is a little slow today. maybe that is it [17:39] jdstrand, oh yeah, right. [17:39] Saviq, want a silo then? [17:40] robru, don't think I have anything that's worth a silo now [17:40] kgunn, unless we want dash-as-app in? [17:40] ↑ [17:40] Saviq, the spreadsheet shows a pending request from you and mhr3 [17:41] robru, does it now! [17:41] Saviq, line 35 [17:42] robru, that's in silo [17:42] oh wait [17:42] hrm [17:43] Saviq, ah, got confused with line 25, which says it's superceded by 35. ok, deleting that [17:43] robru: me too! [17:43] alecu, ok! [17:44] alecu, you want line 33? [17:44] robru: yes, please. And I've got permission from mhr3 to override silo 5 that also holds click scope. [17:50] alecu, boiko: ok please build [17:51] robru: thanks [17:51] boiko, you're welcome! [17:54] cyphermox, ok, you got 17 [17:57] cyphermox, will publish 3 soon, just waiting for some stuff in proposed so I can kick an image [17:57] robru: sure, thanks! [17:58] cyphermox, you're welcome! [17:58] cjwatson: we seem to be having trouble with the screen unlocker now. Going to try switching up the order but I don't suspect it will help [17:59] huh, can't see why that would have changed in stable [17:59] dbus.exceptions.DBusException: org.freedesktop.DBus.Error.AccessDenied: Rejected send message, 1 matched rules; type="method_call", sender=":1.24" (uid=32011 pid=2178 comm="python3 -c import dbus, logging;from unity8 import") interface="com.canonical.powerd" member="requestSysState" error name="(unset)" requested_reply="0" destination=":1.1" (uid=0 pid=899 comm="/usr/bin/powerd ") [17:59] cjwatson: I think we've seen this bug in the unlocker before, let me check to see if it looks the same as the one I filed previously [18:00] cjwatson: the one I've normally seen before is https://bugs.launchpad.net/ubuntu/+source/unity8/+bug/1347183 but maybe it's the same cause? [18:00] Launchpad bug 1347183 in unity8 (Ubuntu) "unity8 unlocker code fails sometimes" [Undecided,Confirmed] [18:01] plars: I know nothing [18:05] bfiller, we're out of silos, you'll have to wait abit [18:05] robru: np, I have a couple hoping to release shortly too [18:06] bfiller, sure, just gotta build an image before I can release anything [18:06] robru, yup [18:06] robru, line 41 could use one [18:06] tvoss, oh yeah, now you respond ;-) [18:07] robru: can i getta silo reconfig on silo14 ? [18:07] robru: and not sure i thanked you y'day for last minute tap dance on qtcomp...so thank you [18:07] kgunn, you're welcome! [18:08] kgunn, also yes [18:27] kgunn: The fix for the poweroff-dialog-on-screen-on already landed, right? [18:28] I just ran into that twice in a row on image 159. [18:31] * sil2100 is back [18:31] Ok, doing teh e-mailz [18:32] ogra_, brendand: thanks [18:34] sil2100, so since 160 had not much changes we wanted to wait for 161 ... apparmor fix is in flight in the archive and robru will roll an image once it is in [18:34] ogra_, sil2100: yep apparmor is in proposed, will kick an image as soon as I see that land [18:36] o/ [18:37] Thanks guys, excellent news that a fix is in [18:37] This would mean that we might have something for tomorrow morning [18:37] i also just uploaded an lxc-android config that should fix audio recordinng [18:38] (sits in proposed as well and should make the next publisher run) [18:39] ToyKeeper: hmmm...guess its just racy... [18:39] can you reopen ? [18:43] kgunn: Looks like it was never closed... but I did update it. [18:45] * ogra_ grumbles [18:45] http://people.canonical.com/~ogra/ubuntu-touch/broken-download.png [18:46] davmor2, around? [18:47] tvoss: I am currently yes [18:47] gatox, ^^^i assume i shouldnt see the dbus error there (it stays if i resume the download and the button keeps saying "resume") [18:48] ogra_, no [18:49] http://people.canonical.com/~ogra/ubuntu-touch/broken-download2.png [18:49] for the record :) [18:49] mterry: doe https://jenkins.qa.ubuntu.com/job/utopic-touch_stable-mako-smoke-daily/6/console look to be similar to that other unlock bug I filed? I can't seem to get past this one by rerunning [18:50] plars, oh!? always reproducable? [18:51] mterry: I've run it 4 times or so on that image, and hitting it every time so far [18:51] mterry: note that this is on the stable-staging-proposed channel image [18:52] plars, I can look at it tomorrow [18:56] robru, hey, can silo 2 be published? [18:56] oSoMoN, yep, once apparmor finishes landing and I kick an image [18:57] robru, ok, thanks. Is there an ETA for the apparmor landing? [18:57] oSoMoN, it's in proposed already, autopkgtests are running. not sure. should be an hour or two I guess. [18:58] autopkgtests are done [18:59] so needs publisher, proposed-migration, publisher [18:59] an hour? [19:00] (p.s. grumble it wasn't this slow not that long ago) [19:00] maybe I shouldn't have given up on my optimise-dominator branch so early [19:02] * ogra_ hopes lxc-android-config makes the same publisher run [19:02] but it looks like it is one cycle behind :/ === fginther changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard (general help): fginther | CI Train support: robru, trainguards | CI Train Status: #157 promoted | Known issues: Both queuebot and http://bit.ly/1mDv1FS know your silo status before the spreadsheet does. [19:22] popey: can you try and play a video from the video scope please I'm getting No video selected to play [19:22] ToyKeeper: ^ [19:27] davmor2: Under what conditions? [19:29] ToyKeeper: just got to the video scope select a video that is on the device from the carousel and then on the details page hit play [19:29] davmor2: Yeah, but what image? It worked fine last time I tried, on 159. [19:29] ToyKeeper: 160 [19:30] Sure, after I update. [19:35] davmor2, ack I see that here [19:36] pmcgowan: I blame ogra who will no doubt blame mterry :) [19:36] its all mterry's fault ! [19:36] see [19:37] ogra_: I'm wondering if it is a side effect of the permissions fix [19:37] what exactly ? [19:37] ogra_: I can see them in the scope but mediaplayer won't play videos saying there are none [19:38] kenvandine, you got silo 20 [19:38] robru, you sure are anxious to get rid of silos :) [19:38] robru, thx! [19:38] kenvandine, you're welcome! [19:38] davmor2, this worked in 159? [19:38] davmor2, well, that cant really be related to the passwd changes ... [19:38] i'm hoping to add a couple more branches to that, if i can get ci passing [19:39] pmcgowan: ToyKeeper said yes I went from 157 to 160 [19:39] kenvandine, branches on the same source project? or different ones? [19:39] at least i couldnt imagine how ... since the passwd stuff obviously works if you can start the session [19:39] 158 had a ton of landings, but not 159 or 160 [19:39] davmor2, did you check syslog for denials ? [19:40] pmcgowan, well, 160 was specifically built for the passwd change ... [19:40] ok [19:40] ogra_: not had time to look anywhere yet busy with bug confirmation for rtm14 tags [19:40] ogra_, what happened? [19:40] locations service and push client just got lucky to hit the archive at the right time ;) [19:40] mterry, seeems davmor2 cant play videos from the scope anymore [19:41] ogra_, apparmor again? [19:41] ogra_, mterry: local videos show in the scope and then you hit play and mediaplayer say it can't there are no videos [19:42] mterry, well, not sure [19:42] Jul 30 15:42:13 ubuntu-phablet kernel: [14261.663293] type=1400 audit(1406749333.369:13404): apparmor="DENIED" operation="open" profile="/usr/bin/mediascanner-service-2.0" name="/var/lib/extrausers/passwd" pid=3153 comm="mediascanner-se" requested_mask="r" denied_mask="r" fsuid=32011 ouid=0 [19:42] that was fixed a while ago [19:43] thats 160 [19:43] unless I ave cruft [19:43] have [19:44] jdstrand, we changed it a bit ... the file is now an actual file again and the dir is a bind mount ... up to 159 the file was a bind mount [19:44] could that have any impact ? [19:45] pmcgowan: apparmor_parser -p /etc/apparmor.d/usr.bin.mediascanner-service-2.0 [19:45] pmcgowan: actually, just this: apparmor_parser -p /etc/apparmor.d/usr.bin.mediascanner-service-2.0 |grep extrauser [19:45] # When using libnss-extrausers, the passwd and group files are merged from [19:45] /var/lib/extrausers/group r, [19:45] /var/lib/extrausers/passwd r, [19:46] same [19:46] ogra_: it shouldn't [19:47] jdstrand, even if an old profile exists where the file was a bind mount ? [19:47] davmor2: Same here, on image 160 videos give me the same error. [19:47] ogra_: wait, what changed? [19:47] I thought we were talking about the libnss-extrausers file [19:48] pmcgowan: can you paste the output of 'stat /etc/apparmor.d/usr.bin.mediascanner-service-2.0 /etc/apparmor.d/cache/usr.bin.mediascanner-service-2.0' [19:48] jdstrand, http://launchpadlibrarian.net/181042722/lxc-android-config_0.178_0.179.diff.gz [19:49] yeah, that shouldn't matter. the path in the denial is the same [19:49] ok [19:49] http://pastebin.ubuntu.com/7907782/ [19:51] pmcgowan: I'm at a loss [19:52] thats not good ! [19:53] let me see something [19:54] pmcgowan: can you do 'tar -zcvf /tmp/aa.tar.gz /etc/apparmor.d' and put aa.tar.gz somewhere? [19:55] jdstrand, email ok? [19:55] pmcgowan: sure [19:57] ogra_, ugh, this is the logest migration EVAR [19:57] yeah [20:01] fginther, can I get a silo for line 46 in the spreadsheet? [20:02] or robru ^ [20:03] jhodapp, nope, none left, sorry [20:03] robru, ok np, it's not critical that I have one today [20:03] robru, I'll check back tomorrow [20:20] slangasek: given the SystemImageInfo in https://errors.ubuntu.com/oops/e3443c10-176b-11e4-ad0f-fa163e5bb1a2 - what do you think we would want a counter for? "version version" or "device name:version version"? [20:22] ugh [20:22] 140 packages of KDE stuff just hit the publisher ... i guess that wont help speeding it up [20:24] ogra_, ah, lp says it's migrated. just waiting for rmadison to catch up with lp [20:24] I mean apparmor [20:24] ah, sad, i was hoping lxc-android-config would make it too [20:26] davmor2, do you see those apparmor denials on mediascanner? [20:26] or ToyKeeper ^^ [20:27] pmcgowan: I get a bunch of those on first boot anyway, because my post-flash script pushes a bunch of media with the wrong permissions. [20:27] I need to modify it to push into /tmp, chown, then move it into /home. [20:27] wait [20:28] ok [20:28] the denials we are interested in are specifically ones for /var/lib/extrausers/passwd [20:28] (instead of pushing into /home then chowning it) [20:28] ToyKeeper's denials are different [20:29] I should have been more specific [20:29] personal denials for everyone ... yay [20:30] I fear the day is coming when I need to wipe [20:30] tyhicks: hey, let's wait for sbeattie [20:30] ack [20:31] the original question is why dont vids play [20:31] jdstrand, seems my issue is separate from just playing videos [20:31] pmcgowan, the media-player log should reveal that ... or the url-dispatcher [20:31] pmcgowan: The videos get thumbnailed just fine, but the media player doesn't play them. [20:32] tyhicks, sbeattie: hey, so pmcgowan has a bizarre situation where the policy for usr.bin.mediascanner-service-2.0 has '/var/lib/extrausers/passwd r,' in it, but is still getting a denial [20:33] pmcgowan: I see something odd with music too. The scope preview can play them, but the music app launches as if no file were specified. [20:33] tyhicks, sbeattie: this happens after verifying the profile with: [20:33] $ apparmor_parser -p /etc/apparmor.d/usr.bin.mediascanner-service-2.0 |grep extrauser [20:33] bdmurray: we would want counters for version ubuntu, and version version [20:33] # When using libnss-extrausers, the passwd and group files are merged from [20:33] /var/lib/extrausers/group r, [20:33] /var/lib/extrausers/passwd r, [20:33] pmcgowan: So, I suspect it's not a permission issue at all, but rather a matter of the scope passing the filename to the app. [20:33] tyhicks, sbeattie: running apparmor_parser -r /etc/apparmor.d/usr.bin.mediascanner-service-2.0 didn't help [20:33] bdmurray: oh, sorry - I mean, 'version ubuntu' and 'device name:version version' [20:34] ToyKeeper, but that stuff didnt change in the image that we know of [20:34] pmcgowan: The music player can play normally if I select a song from within the app, but the video player doesn't have that functionality. [20:34] tyhicks, sbeattie: the denial is: Jul 30 15:42:13 ubuntu-phablet kernel: [14261.663293] type=1400 audit(1406749333.369:13404): apparmor="DENIED" operation="open" profile="/usr/bin/mediascanner-service-2.0" name="/var/lib/extrausers/passwd" pid=3153 comm="mediascanner-se" requested_mask="r" denied_mask="r" fsuid=32011 ouid=0 [20:34] bdmurray: because 'version version' alone tells you the current build number on the channel but doesn't tell you which channel [20:34] tyhicks, sbeattie: let me forward you the tarball of his /etc/apparmor.d [20:34] bdmurray: so we want 'version ubuntu', for "which rootfs build is this"; and 'device name:version version', for "which device image is this" [20:35] tyhicks, sbeattie: emailed [20:36] * sbeattie looks [20:36] slangasek, probably not much different from what we use for "build number" on the dashboard :) http://ci.ubuntu.com/smokeng/utopic/touch/ [20:36] ogra_: well, the point is that we want to be able to correlate crashes across devices, and within a single device [20:37] so we need two sets of counters [20:37] yeah [20:37] tyhicks, sbeattie: I checked the stat of the files: http://pastebin.ubuntu.com/7907782/ [20:37] jdstrand: have you tried to reproduce it? [20:37] tyhicks, sbeattie: no one else sees it [20:38] tyhicks, sbeattie: we had to apparmor_parser -R to get it to stop cause the mediascanner kept restarting cause it would crash on the failure [20:38] jdstrand, pmcgowan: what device and kernel version? [20:38] tyhicks, Linux ubuntu-phablet 3.4.0-5-mako #32-Ubuntu SMP PREEMPT Wed Jul 16 10:39:52 UTC 2014 armv7l armv7l armv7l GNU/Linux [20:39] jdstrand, I haven't wiped in some time, perhaps I have some bit rot [20:39] jdstrand, dont want to start a goose chase [20:39] pmcgowan: well, the thing is, this is very, very odd [20:39] but that would be buggy bit rot [20:39] yeah [20:39] as in, this should never happen [20:39] I asked tyhicks and sbeattie in case I am overlooking something [20:40] tyhicks, sbeattie: I wonder if it has something to do with profile replacement [20:40] slangasek: ack [20:41] tyhicks, sbeattie: eg, if we loaded the profile now (or had Pat reboot), and see if the errors are still there [20:41] s/errors/denials/ [20:41] jdstrand: rebooting would be the most complete test of that [20:42] pmcgowan: can you reboot? when done, can you see if there are denials post reboot and also paste 'sudo aa-status'? [20:42] jdstrand, yep [20:43] jdstrand: if that doesn't fix it, I'd like to skip the cached binary profile (or just purge the cache) when doing a reload of the profile [20:43] I think -T option to apparmor_parser would do that [20:46] and/or --show-cache so we see what the parser thinks of the cache state [20:47] jdstrand, tyhicks no denials, and I don't see mediascanner running which is probably correct [20:47] http://pastebin.ubuntu.com/7908231/ [20:48] pmcgowan: can you add a video to the ~/Videos folder, with mtp? (or just make sure it is copied in place as the phablet user) [20:48] robru: oops, silo-10 now has all it's MPs approved. What needs to happen for it to publish? :-) [20:49] robru, what’s the status of the apparmor publication and the new image you were going to kick? [20:50] sbeattie: where I was going with the -T option was to see if the cached binary was corrupted at all - maybe a few bits were flipped in that particular pathname or in that particular access bit field [20:50] jdstrand, no denials but I don't see the scanner kick in [20:51] pmcgowan: not sure how to get it to start [20:51] pmcgowan: maybe got to the Videos scope? [20:51] it should with the folder change [20:51] I did [20:52] know here that log file is? [20:52] /home/phablet/.cache/upstart/*mediascanner*log (though, you might want to expand the glob to make sure you snag the right file) [20:53] pmcgowan: /home/phablet/.cache/upstart/mediascanner-2.0.log it looks like [20:53] jdstrand, odd its not starting [20:54] wtf [20:54] jhodapp: fyi, apparmor-easyprof-ubuntu 1.2.14 uploaded for your micshm path change [20:54] jdstrand, ogra_ just mentioned he cant see new videos, we have several issues here [20:55] pmcgowan: what does the log say? [20:56] jdstrand, now video is playing [20:56] jdstrand, the log was never opened, only old .gzs [20:56] looking at my phone, it seems mediascanner service should be a long running process [20:56] pmcgowan: ah, so mediascanner is running now? [20:56] jdstrand, no [20:57] but mediaplayer succeeded to play the video [20:57] I cant add new ones, but can play old ones [20:57] pmcgowan: did you modify its upstart job or have a silo enabled or something? [20:58] jdstrand, nope [20:58] slangasek: would we want to query on "device image" and "ubuntu release" or would you consider those separate? [20:58] pmcgowan: start mediascanner-2.0 [20:58] pmcgowan: do that as the phablet user [20:58] jdstrand, i can do a flash and see if anything gets better [20:58] ok [20:59] I would expect that to generate the log file [20:59] its crashing again [20:59] pmcgowan: with apparmor denials? [20:59] yep [20:59] bdmurray: hmmm we do reset the channel numbers at the start of the release, I think, in which case the release needs to also be part of it [20:59] stgraber: ^^ can you confirm? [21:00] tyhicks: do you want to take over here with the alternate options? [21:00] jdstrand, terminate called after throwing an instance of 'std::runtime_error' [21:00] what(): Could not look up user name: Permission denied [21:00] from mediascanner log [21:01] pmcgowan: how about 'sudo apparmor_parser -rT /etc/apparmor.d/usr.bin.mediascanner-service-2.0 [21:01] pmcgowan: that will reload the profile without the cache file [21:01] ah, we're back to crashing [21:01] good [21:02] tyhicks: it didn't start on boot, so I had him start it manually via its upstart job [21:02] ok [21:02] which showed the denials [21:02] jdstrand, that fixed things up it seems [21:02] not crashing, no denials [21:02] very interesting [21:02] yeah [21:02] * sbeattie hrms [21:03] tyhicks, sbeattie: could this be fallout from the caching patches the we have? [21:04] * tyhicks is thinking [21:05] pmcgowan: ok, last thing. can you do: rm -f /etc/apparmor.d/cache/usr.bin.mediascanner-service-2.0, then reboot? [21:05] jdstrand: so you *just* changed that profile in your last apparmor upload (a couple days ago) and the cached binary didn't get updated [21:05] tyhicks: well, the stat I did showed that the cache was regenerated on 4/28, the day of the upload [21:06] tyhicks: I changed the nameservice abstraction. I didn't add any other patches [21:07] tyhicks: https://launchpad.net/ubuntu/+source/apparmor/2.8.96~2541-0ubuntu2 [21:07] jdstrand: and that change had already landed in the image that pmcgowan was running? [21:09] jdstrand, tyhicks: regenerating the cache file locally using the parser as built from the 2.8.96~2541-0ubuntu2 generates a slightly different cache file than what was in p mcgowan's tarball [21:10] jdstrand, rebooted [21:10] tyhicks: yes, pmcgowan is running the latest image. I verified that 'apparmor_parser -p /etc/apparmor.d/usr.bin.mediascanner-service-2.0 | grep extrausers' had what I expected first [21:11] sbeattie, jdstrand: I'm wondering if his image had the usr.bin.mediascanner-service-2.0 cached binary from 2.8.96~2541-0ubuntu1 [21:12] seems quiet [21:12] tyhicks, sbeattie: so in case you weren't aware, jj had all those patches to make it so you could run the parser against an apparmor features dir, and rsalveti starting using that to generate server cache files recently [21:12] if that was the case, it would be missing the '/var/lib/extrausers/passwd r,' rule [21:12] scanner running and no denials [21:12] tyhicks: why wouldn't it have updated on boot? [21:13] tyhicks: yeah that was what I wanted to verify in my question to #security [21:13] jdstrand: I don't know for sure, but possibly from timestamp oddness between the text profile and the binary profile [21:14] tyhicks: your theory seems to go with what sbeattie mentioned about the cache being slightly different [21:14] (it would be a 2 rule difference) [21:14] rsalveti: when you generate the apparmor profile cache for the system images, where do you pull the profiles from? [21:15] jdstrand: well, oddly, my locally compiled cache file is slightly smaller than the one in the tarball. [21:15] jdstrand: tyhicks: we're not copying the cached files for the ones available in /etc/apparmor.d/cache [21:15] sbeattie: heh [21:15] when updating the image [21:15] pmcgowan: now that you've rebooted, can you paste the output of 'stat /etc/apparmor.d/cache/usr.bin.mediascanner-service-2.0' [21:15] because the currently logic in the initrd that copies over the files doesn't replace them if they are already there [21:16] rsalveti: jdstrand recently uploaded apparmor 2.8.96~2541-0ubuntu2, which only made a small policy change, and it looks like the policy from 2.8.96~2541-0ubuntu1 may be what was cached [21:16] but, they should still be updated by apparmor during boot [21:16] if not, that's a bug [21:16] sbeattie, http://pastebin.ubuntu.com/7908468/ [21:17] rsalveti: hrm... ok, we'll look into why the parser didn't catch the new profile [21:18] pmcgowan: thanks. it does show there's a difference in what was in the cache previously (Size: 78713) versus the working version (Size: 78689) [21:20] tyhicks, sbeattie: so, I don't think two people necessarily need to work on it. can you guys work out who should look at it? (my preference is for sbeattie to keep plugging away at abstract sockets) [21:20] tyhicks, sbeattie: but I appreciate both of you helping to triage it [21:20] seems my work here is done [21:21] thanks pmcgowan [21:21] pmcgowan: you device should be ok now. we'll need to do some investigating [21:21] good stuff [21:21] jdstrand: it probably makes sense for me to look at it anyways, since I've been digging around this profile caching code for the last week or so [21:21] sbeattie: is that ok with you? ^ [21:21] tyhicks: fine by me. [21:22] rsalveti: so, to be super clear, when you run the serverside stuff, you will pull in /etc/apparmor.d from the latest apparmor to make sure it has the latest policy? [21:22] tyhicks, sbeattie: thanks again guys :) [21:24] jdstrand: the problem is that we're not always copying the files when booting the system [21:24] === trainguard: IMAGE 161 building (started: 20140730 21:25) === [21:25] the logic in the initrd first check if the dest file is already there, if so, don't overwrite it [21:25] even if the content is different [21:25] that's not a problem for click packages as they are named with the version id [21:25] for the files available in /etc/apparmor.d, that's indeed an issue [21:25] so we ended up not updating the cache files if they are already there, with an older version [21:26] and we just let apparmor to recreate such cache during boot [21:26] oh [21:26] the problem here is that for some reason apparmor is not updating the cache when booting the system [21:26] I misunderstood earlier [21:27] and still trying to use the older version [21:27] ah [21:27] rsalveti: so, if the cache file is newer than the system policy file, then it won't update [21:28] cause we aren't forcing a cache regeneration [21:28] even though the policy actually changed in an abstraction [21:29] * jdstrand looks at the upstart job [21:29] jdstrand: but in this case the cache file is older, isn't it? [21:29] jdstrand: the parser should probably be checking the timestamps of abstractions when deciding if it needs to update the cache file [21:29] on Pat's system, no, it wasn't [21:30] that doesn't make much sense then [21:30] tyhicks: yeah. you see what I am saying, right? [21:30] if it's newer, then it got automatically generated during boot [21:30] rsalveti: I think it does [21:30] * sbeattie wonders what a stat of the abstraction looked like in comparison [21:30] rsalveti: ie, sometime ago, he got a new mediascanner-2.0 profile [21:30] rsalveti: it updated the cache [21:30] right [21:30] then, he did an image update [21:31] the abstraction changed [21:31] so the timestamp on the profile didn't [21:31] oh, got it [21:31] kind of a race [21:31] eh, more of a buggy parser [21:31] the upstart job has logic to attempt to deal with that [21:31] (in the absense of the parser doing the right thing) [21:31] it should be checking all timestamps of #include'ed files [21:32] lets just make the parser do the right thing [21:32] oh sigh [21:32] the upstart job logic only deals with this for clicks [21:32] hmm [21:32] tyhicks: do you see it? [21:33] if [ -n "$run" ]; then [21:33] aa-clickhook -f [21:33] fi [21:33] I haven't looked at the upstart job [21:33] tyhicks: if apparmor changed, we need to force a cache regeneration [21:33] (and run aa-clickhook -f) [21:34] tyhicks: do you have it handy? [21:34] jdstrand: I do - looking at it now [21:34] tyhicks: it is the [ -x /usr/bin/aa-clickhook ] hunk [21:35] tyhicks: it is wrong for how the server side is handling precaching for system policy [21:35] tyhicks: cause it only cares about click and assume system policy will just work [21:37] jdstrand: but if the parser was fixed to check the timestamps of all of the #include'ed files, couldn't that entire hunk go away? [21:37] tyhicks: the 'apparmor' part could [21:37] I thought the parser did that, but could be mis-remembering [21:37] tyhicks: not the apparmor-easyprof-ubuntu part [21:38] also, in this case, it still would've avoided the cache [21:38] sbeattie: I'm blindly assuming that it doesn't - I'll verify in a moment [21:38] err, I mean, used the cace [21:38] s/cace/cache [21:38] sbeattie: why? was the abstraction's timestamp not newer than the cache file? [21:38] (I thought we didn't know the abstraction's timestamp) [21:39] the mtime is about an hour before the cache file [21:39] $ stat ~/tmp/etc/apparmor.d/cache/usr.bin.mediascanner-service-2.0 | grep ^Modify [21:39] Modify: 2014-07-28 08:28:53.000000000 -0700 [21:39] $ stat ~/tmp/etc/apparmor.d/abstractions/nameservice | grep ^Modify [21:39] Modify: 2014-07-28 07:54:05.000000000 -0700 [21:39] tyhicks: we have to for apparmor-easyprof-ubuntu because we need to regenerate the policy files from templates [21:40] jdstrand: oh, right! [21:40] slangasek: haven't looked at the whole backlog, but yes, each series has its own channel and when the channel is created it starts at buildid 1 [21:41] sbeattie: how did that happen? [21:41] slangasek: then we change the target of the devel-* aliases to the new series which basically means the devel-* channels appear to rollback from buildid 100 or so down to 1 [21:41] slangasek: the system-image client tracks both the buildid and the channel alias target so that if the target changes it assumes its internal buildid to be 0 and forces a full update to the new build [21:42] tyhicks: presumably the apaprmor cache update occurred locally on pat's phone in between the time the ubuntu2 package was built and him applying the update. [21:43] tyhicks, sbeattie: could he have booted before running a system-image update (therefore, had a cache from that day), then did a system-image update which updated the abstraction? [21:44] sbeattie: that would explain why we don't see more widespread issues-- it had to be timed right [21:44] or rather, wrong :) [21:44] yeah [21:44] man, that's a pain [21:45] brendand, I think I have the job fixed that was causing your address-book-service jobs to fail. I've retriggered a build to the results [21:46] tyhicks: so, wouldn't updating the upstart job to also force regenerate system policy like we do for click policy fix this? [21:46] fginther, i didn't even know you were looking at it - thanks :) [21:47] brendand, om26er mentioned it to me earlier today === fginther changed the topic of #ubuntu-ci-eng to: Ubuntu CI Engineering Team | Vanguard (general help): cihelp | CI Train support: robru, trainguards | CI Train Status: #157 promoted | Known issues: Both queuebot and http://bit.ly/1mDv1FS know your silo status before the spreadsheet does. [21:47] tyhicks: ie, if the apparmor md5sum changed since last time, regenerate? [21:48] tyhicks: (and by force regenerate, I mean removed /etc/apparmor.d/cache/* [21:48] ) [21:48] remove* [21:49] jdstrand: yeah - that would do it [21:49] fginther, did you find a solution for the otto issue ? [21:49] jdstrand: we'd probably need to use clear_cache() from /lib/apparmor/functions, so we'd need to do that after sourcing in that file [21:49] om26er, yes :-) [21:49] tyhicks: right [21:50] fginther, has it been "deployed" ? [21:50] tyhicks: can you add that to your list of tings to update in the next apparmor upload? [21:50] jdstrand: just did that :) [21:50] om26er, yes, I was in the middle of finding the build you sent me so that I can re-start it [21:50] tyhicks: I'm happy to review the upstart job changes if you want [21:50] tyhicks: thanks! [21:50] ok, nice. [21:50] mystery solved [21:50] jdstrand: yeah, I'll def want you to review it [21:51] fginther, https://jenkins.qa.ubuntu.com/job/autopilot-testrunner-otto-utopic/1824/console [21:51] man, who knew that md5sums stupid hack had legs [21:51] om26er, ahh sweet. I already restarted that one [21:51] aka, caching is hard :) [21:53] fginther, can i rerun the job then? [21:53] brendand, already did [21:54] brendand, om26er, it's s-jenkins.ubuntu-ci:8080/job/address-book-service-ci/274/ [21:54] fginther, cool. [21:54] I see it did pass in otto after your change http://s-jenkins.ubuntu-ci:8080/job/autopilot-testrunner-otto-utopic/1834/console [21:58] ogra_ (and anyone else who was paying attention): pmcgowan's libnss-extrausers denial is triaged. This is not a widespread problem (no bugs) since it relied on a very specific timing of events. that said, we know the issue and will fix it in the next apparmor upload, scheduled for some days out [21:59] tyhicks: can you file a bug for that? ^ [22:06] jdstrand: ack - will do [22:17] tyhicks: thanks! :) [22:54] === trainguard: IMAGE 161 DONE (finished: 20140730 22:55) === [22:54] === changelog: http://people.canonical.com/~ogra/touch-image-stats/161.changes ===