[08:54] davmor2, Hey, what are the critical bugs on your list for the release of vivid? [08:57] jibel: so it turns out that somehow zsync and mangled the images. Checksums were well off. So I killed them added them fresh. Wifi is still playing up, not rebooting on hitting enter, and oem mode is still leaving the temporary user I'll grab you the bug numbers in a second [09:00] davmor2, shutdown is known, OEM I'm double-checking it was supposed to be fixed. [09:01] jibel: might of been over the weekend maybe but certainly wasn't on Friday when I got the right image [09:01] davmor2, we are looking for someone from design for the vivid slideshow [09:02] davmor2, it was fixed 2 weeks ago [09:02] davmor2, so apparently the fix is incomplete [09:02] jibel: they are in the office just lock the door around lunch and put on a sign that says you can't eat till someone gives me the slide designs ;) [09:03] jibel: I'll check the yelp too and see if the artwork there is updated yet [09:04] jibel: bug 1445587 [09:04] bug 1445587 in ubiquity (Ubuntu) "Installer hangs after clicking "Restart Now" when running under qemu" [High,In progress] https://launchpad.net/bugs/1445587 [09:04] pitti, bug 1445592 [09:04] bug 1445592 in ubiquity (Ubuntu) "vivid fails to shutdown or reboot at end of installation when "reboot now" button is pressed" [High,Confirmed] https://launchpad.net/bugs/1445592 [09:33] jibel: https://bugs.launchpad.net/ubuntu/+source/network-manager/+bug/1445481 [09:33] Ubuntu bug 1445481 in network-manager (Ubuntu) "My ap is missing from OEM desktop mode but was found no issues during the install" [Undecided,New] [09:34] jibel: ^ that one bites me from time to time [09:34] jibel, davmor2 I can confirm the removal of the OEM user is still an issue - https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/1442679 [09:34] Ubuntu bug 1436937 in ubiquity (Ubuntu Vivid) "duplicate for #1442679 Temporary OEM user not removed after end user setup" [High,Triaged] [09:35] jibel: ^ that's the other one I'm looking for :) [09:35] flexiondotorg: thanks [09:35] davmor2, flexiondotorg on it [09:35] still don't know what is happening there [09:44] jibel: I'm assuming it is a final step of an upstart job that hasn't been ported to systemd, so just blame pitti ;) [09:44] davmor2: yes, please do; I'm on it [09:44] ah, actally I'm on bug 1445592 for now [09:44] bug 1445587 in ubiquity (Ubuntu Vivid) "duplicate for #1445592 Installer hangs after clicking "Restart Now" when running under qemu" [High,In progress] https://launchpad.net/bugs/1445587 [09:45] it's likely that the OEM user bug is mine, too, but of course any help appreciated [09:45] pitti: Yeah I'll be installing again, is there any info I can drag off the system that would be useful to you? [09:48] davmor2: I suppose this bug shold be fairly easy to reproduce, so as long as we have a bug which clearly documents what's wrong and how to get there I'm happy [09:48] easier to reproduce locally than ssh'ing to a remote one [09:49] pitti: I think jibel is already testing it [10:05] davmor2, jibel I also confirm this - https://bugs.launchpad.net/ecryptfs-utils/+bug/1445198 [10:05] Ubuntu bug 1445198 in ubuntu-mate "Failed script for encrypt home after installation" [Undecided,Confirmed] [10:05] davmor2, jibel I suspect this affects all flavours. [10:12] flexiondotorg, thanks, I'll look into it [10:15] jibel, Thanks. [10:40] jibel: I'm grabbing netboot and starting there [10:40] davmor2, thanks [10:41] I'm trying to reproduce the ecryptfs issue, it seems it is not affecting ubuntu [10:42] davmor2, when you'll run an install on HW, can you click on the 'release notes' link and tell if it works [10:42] jibel: will do [10:42] davmor2, in ubiquity-dm [10:43] jibel: yeap [10:47] jibel, I confirm the ecryptfs issue doesn't affect Ubuntu. [10:47] jibel, It did used to work in Ubuntu MATE. [10:47] jibel, How is that script being invoked? [10:47] flexiondotorg, right, there is something missing from Mate, that's what I'm checking [10:48] jibel, I suspect xdg-open is missing MATE support. [10:48] jibel, Can you point me at the code that calls the script? [10:49] https://bugs.launchpad.net/ubuntu-mate/+bug/1001902 [10:49] Ubuntu bug 1001902 in ubuntu-mate "xdg-open is not mate compatible" [Medium,Confirmed] [10:54] flexiondotorg, the problem is that xterminal-emulator doesn't support double-quotes to execute shell commands [10:55] jibel, Ah. [10:55] jibel, So why is Ubuntu not affected? [10:55] flexiondotorg, for example x-terminal-emulator -e sh -c 'echo hello && sleep 5' will work [10:56] for xterminal-emulator -e "sh -c 'echo hello && sleep 5'" won't [10:56] s/for/but/ [10:58] jibel, Thanks for the info. [11:02] flexiondotorg, 2 options: change mate-terminal to support double-qoutes since -e is a command, or fix update-notifier hook to execute a command in move the current command executed in the hook to the script. [11:05] flexiondotorg, actually the bug occurs only if you invoke mate-terminal through x-terminal-emulator [11:07] flexiondotorg, the problem is probably in mate-terminal.wrapper [11:08] jibel, Please can you point me at the code that is invoking this? [11:08] flexiondotorg, it's in /usr/bin/mate-terminal.wrapper [11:08] jibel, No :) [11:09] jibel, The update-notifer stuff . [11:11] flexiondotorg, ah, the hook is /usr/share/ecryptfs-utils/ecryptfs-record-passphrase from ecryptfs-utils and linked into /var/lib/update-notifier/user.d/ to trigger the notification [11:11] flexiondotorg, but the bug is in the wrapper I think [11:12] jibel, I'm just curious as to why this had been working fine. [11:14] no idea [11:14] jibel, Thanks. I'll have a dig. [11:14] yw [11:29] jibel, I tracked the behaviour change down to the mate-terminal package. A few months back we changed from mate-terminal being the default to mate-terminal.wrapper. [11:31] flexiondotorg, in the wrapper you probably need to split oldargs to make it a list but don't split the quoted part of the command [11:32] a shell wrapper would be easier actually [12:44] jibel: do you want server tests running on netboot aswell as desktop? [12:46] davmor2, wait there is a new kernel and a new debian-installer has just been uploaded [12:46] shuggin fashin' dick dastardly === plars is now known as plars- [13:13] pitti: so are we right is the oem removal just a missing systemd job? [13:14] davmor2: could be, I didn't look at it at all yet [13:15] davmor2: there is an oem-config.service which ought to do the same as the upstart job (calling oem-config-firstboot) [13:16] pitti: right oem end user is setup, it's just the temporary oem user isn't removed and it auto starts that user :( [13:16] ………………………………………………………………userdel --force --remove oem || true [13:16] I see that in oem-config-firstboot, so perhaps that fails? [13:16] there might still be processes left from the oem user [13:17] although it has a "pkill -u oem || true" right above it [13:31] jibel: https://code.launchpad.net/ubuntu-seeds [13:31] jibel: Release notes links to www.ubuntu.com/index_roadshow but that I'm assuming is because the page doesn't exist yet right? [13:31] davmor2, pitti - cyphermox has wored on the oem-config issue a few weeks back. [13:32] davmor2, it is a redirection server-side. [13:32] davmor2, pitti - cyphermox did add the required systemd support, AFAIK. [13:32] davmor2, pitti - The current issue is that the 'oem' user account is not always (as in rarely) removed. [13:32] flexiondotorg, it is on release team's radar [13:33] davmor2, pitti - The result being that after a system is prepared and the oem config is completed you are auto logged in as the 'oem' user. [13:33] jibel: on a plus side it does open a browser and goto the above link [13:33] davmor2, thanks, in the test I did, it crashed the VM [13:33] jibel, Thanks. Just thought I share. I've been test the oem-config stuff for week. I was working for a brief period. [13:34] *it was working [13:34] flexiondotorg, davmor2: is that systemd specific? I. e. if you boot the first time with init=/sbin/upstart (use -snapshot!) does it work? [13:34] flexiondotorg, np, thanks. [13:34] after running oem-config-prepare and rebooting, there shouldn't even be starting anything as oem [13:35] pitti, I'm not sure what you asking? [13:35] flexiondotorg: if the failure to remove the oem user also happens under upstart [13:35] cyphermox, Correct. the oem-config is running as super user. [13:35] i. e. which direction we should debug this [13:35] pitti, I'll test it with upstart. [13:36] or maybe the oem-config thing has a log which might contain the error message from userdel? [13:36] pitti, Do I need any additional packages installed? [13:36] flexiondotorg: no [13:36] pitti, Thanks. [13:36] flexiondotorg: so I meant, normally you'd do a qemu install in OEM mode, and then you always boot the installation with -snapshot so that you can do it arbitrarily many times witohut having to reinstall [13:37] flexiondotorg: the big issue I see is that I still haven't been able to reproduce the bug you see [13:37] pitti, Understood. [13:37] cyphermox, I can reproduce everytime in VirtualBox and about half the time on laptop at home. [13:37] pitti, there is nothing in the log [13:37] nothing useful [13:37] cyphermox, The OEM I'm working with can only reproduce this on Broadwell NUC and Ultrabook. [13:38] flexiondotorg: only ever with ubuntu-mate or other images as well? [13:38] cyphermox, Ubuntu and Ubuntu mATE. [13:41] pitti, Do you want to install completed with upstart or just the first boot after the system is prepared? [13:41] flexiondotorg: the actual install sholdn't make a difference [13:41] pitti, OK. [13:41] flexiondotorg: so if you still have the orginal unmodified instal, or want to rnu oem-config-prepare again, you can use that [13:41] pitti, Understood. [13:41] flexiondotorg: i. e. we need to decide between "the .service runs at the wrong time" vs. "we broke oem-setup script" [13:42] * flexiondotorg is setting up a test environment. [13:42] pitti: I install on hardware snapshotting it a little harder there ;) [13:42] * pitti introduces davmor2 to the wonderful world of VMs :) [13:43] pitti: vm's are wonderful but if you only test on vm you don't know if things like real drivers work :D Not often you get an nvidia vm ;) [13:44] right, but stuff like the above is easier to test in VMs [13:44] pitti: also uefi + secure boot always works better on a device too :) [13:44] yeah, uefi in qemu works, but not secureboot [13:45] pitti: wow that's impressive you got it to work in qemu it still failed for me :( Vb's worked nicely though [13:45] pitti: it's not running at the wrong time; it's now running as the only thing in its own target [13:46] pitti: so the likeliest scenario is still that one oem-config gets removed, systemd decides to go and bring up the rest of the system, before all the scripts are done running [13:46] Greetings folks! [13:54] cyphermox: oh that sounds like fun fun or not === pgraner-afk is now known as pgraner [14:00] cyphermox: that's not very plausible [14:01] cyphermox: it pkills all oem processes before it userdels [14:01] cyphermox: so we'd have that session been torn down, and/or userdel errors because of "busy user" [14:01] cyphermox: it's more likely that the cleanup code isn't run at all, jibel said that we don't get any userdel errors in the logs [14:01] well, who tells you it's not starting other things at the same time? [14:01] it could [14:01] but that doens't magically stop oem-setup [14:01] that's *exactly* what was going on before I made it its own target [14:02] hence why I'm also kind of surprised it would happen again, I tested this a lot and couldn't get it to fail again [14:02] well, let's just get an image, boot it with -snapshot, and set -x that script to see what's going on [14:02] but it could just be a matter of timing/speed now [14:02] little point in speculating [14:03] pitti, trying that right now [14:03] it seemed to work well enough when I ported it, so obviously something changed [14:06] rhuddie, hey there... can you point me to an ap/python-example explaining the correct use of scenarios? === chihchun is now known as chihchun_afk [14:08] rhuddie, atm I'm looking at test_action_latency.py as guide... just wondering if that's good enough [14:08] MacSlow, hey. sure, I'll get back to you shortly. [14:14] pitti, If I try to use init=/sbin/upstart after the system is prepared oem-config doesn't not run at all. I get automatically logged in to an Ubuntu desktop as the 'oem' user. [14:14] I tested this with Ubuntu proper. [14:14] flexiondotorg: oh, fun [14:15] cyphermox, interesting, if I start oem-config-firstboot with --debug, the oem user is deleted [14:16] jibel, How did you invoke that via kernel options? [14:16] flexiondotorg, I changed the systemd job [14:17] and forced debug there [14:17] jibel: so there's the fix then ;) [14:17] jibel, Thanks. [14:22] pitti, cyphermox jibel Just tested twice with systemd and Ubuntu and no debug. One worked, one did not. [14:23] Will try the --debug to see if it consistently works. [14:23] flexiondotorg: don't [14:23] why is it things always work when you turn on debugging it's like the jinx from hell for qa :D [14:23] flexiondotorg: add set -x please [14:23] --debug changes teh behaviour of that script [14:23] pitti, To what? [14:23] flexiondotorg: to the scrip which does the userdel [14:23] pitti, Understood. [14:23] but jibel is on that ATM [14:23] we want to see a set -x in the failed case [14:23] pitti, I'll wait then. [14:24] jibel, The give Ubuntu MATE a test. I seems to fail a little more often than Ubuntu. [14:28] jibel: yelp art-work is updated I think it look more like a Viking than a monkey though [14:29] davmor2, hehe [14:29] infinity, ^ :) [14:36] davmor2: It's not a viking, it's Wolverine. [14:37] infinity: but that doesn't start with V :P [14:37] infinity: maybe the next release will be Wicked Wolverine and then they can keep the mascot ;) [14:38] davmor2: wascawwy wabbit. [14:39] infinity: We are going after the gaming community with the next release Worlda Warcraft release [14:43] MacSlow, for scenarios, you can see an example here: http://bazaar.launchpad.net/~phablet-team/address-book-app/trunk/view/head:/tests/autopilot/address_book_app/tests/test_delete_contact.py [14:43] rhuddie, thx... will take a look [14:43] MacSlow, you may also have seen the documentation: https://pypi.python.org/pypi/testscenarios/ [14:43] rhuddie, ok [14:55] jibel, infinity: meh empty wastebin is still opening a nautilus window :( [15:12] pitti, jibel I got oem-config to fail on Ubuntu with -x set. Where should I be looking for error logs? [15:14] jibel, That issue with the wastebin. Was that in the live session? - https://bugs.launchpad.net/ubuntu/+source/gvfs/+bug/1445622 [15:14] Ubuntu bug 1445622 in gvfs (Ubuntu) "Trash directory unable to be found" [High,Confirmed] [15:14] flexiondotorg, cyphermox: jibel found something, he'll update the bug [15:14] pitti, Thanks. [15:15] cyphermox: I finally understand bug 1445587 *phew* [15:16] bug 1445587 in ubiquity (Ubuntu Vivid) "Installer hangs after clicking "Restart Now" when running under qemu" [High,In progress] https://launchpad.net/bugs/1445587 [15:20] pitti: ah, great [15:21] pitti: I think it should still be working as a proper dm, but I guess we can defer those changes to post-V [15:22] cyphermox: I made it work with a quick hack locally (defer the reboot by 5 s) [15:22] cyphermox: now seeing how we can fix this properly to avoid the race [15:25] cyphermox, added more info to 1436937, userdel fails because it cannot open /etc/subuid [15:29] jibel: if you haven't scrapped this yet, what was running as process 706? [15:30] while oem-config is its own "runlevel", I'm not sure what could have been started running as oem [15:34] cyphermox, systemd [15:34] ^ [15:34] wha? [15:34] pitti, ^ [15:34] pitti: ^ [15:34] :) [15:35] no idea what that could be besides maybe a user session? [15:35] (target != runlevel, FTR) [15:36] pitti: inasmuch as it's the default target, and IIRC doesn't start multi-user OR graphical, for all intents and purposes, there should be nothing else running [15:37] pitti: feel free to check debian/ubiquity.target to see if I messed up [15:37] cyphermox: from what I see on jibel's machine this isn't related to service startup at all [15:37] cyphermox: one can re-run that userdel, and it keeps failing in exactly the same way, stumbling over /etc/subuid [15:37] then how do you explain any process starting as 706? [15:37] I mean, starting as the oem user [15:38] cyphermox: I don't know, does it? but that's not even why it fails [15:38] the expectations for oem-config are simple: nothing else should be running, especially not something as the oem user, and if something does and can't be killed for whatever reason, of course userdel will fail [15:39] it is apparently not failing because there's something running [15:39] if systemd wasn't running as oem, you wouldn't have trouble with /etc/subuid. [15:40] ah, I see jibel's log [15:40] he still has the VM, but he's in a meeting ATM [15:40] in any case, since you seem to understand all this better than I do, I can fix console-setup instead. [15:40] cyphermox: the debconf question? yeah, that'd be appreciated [15:40] yeah [17:11] cyphermox: so the oem user remove one is a really funny bug [17:12] cyphermox: wgrant just figured it out [17:12] (bug in shadow) [17:13] ok [17:21] wxl, assuming you are still kosher with it, I'm going to add the lubuntu lxqt session to the show and tell track for uos 15.05 [17:22] balloons: yeah we'll figure it out :) [17:22] wxl, got a cool session title? [17:22] or preferred day / time? [17:22] ummmmmmmmmmm [17:22] "lubuntu, cuter" [17:22] :) [17:23] and sensible hours given the fact that i'm not utc :) [17:25] wxl, ok I'm make it late in the day [17:26] balloons: so yeah afternoon here is good :) [17:27] wxl, http://summit.ubuntu.com/uos-1505/meeting/22422/lxqt-next-generation-of-lubuntu/. Feel free to change the description or title :-) [17:27] thx balloons [18:23] https://bugs.launchpad.net/ubuntu/+source/nautilus/+bug/1445595 [18:23] Ubuntu bug 1445595 in nautilus (Ubuntu) "Empty Trash Results in File Windows Opening" [Low,Confirmed] [18:24] anyone available to triage that bug [20:06] veebers: hey, renato made a good comment in here: https://code.launchpad.net/~canonical-platform-qa/address-book-app/fix_base_class/+merge/256243 [20:07] with the recommended way of launching the app, then we require to import the AddressBookApp custom proxy object, even if we don't use it. [20:07] or well, we don't use it directly. [20:08] it's one of the problems I see with the registry, we need the import which is not an explicit way of stating what we want. [21:02] balloons, ping? [21:03] knome, pong [21:04] this can't be right: [21:04] For a normal to normal upgrade, terminal will show Prompt=normal [21:04] For a normal to LTS upgrade, terminal will show Prompt=lts [21:04] For a lts to normal upgrade, terminal will show Prompt=normal [21:04] For a lts to lts upgrade, terminal will show Prompt=normal [21:04] balloons, shouldn't everything but the last show normal, and the last one LTS? [21:05] knome, I remember going through this with elfy actually [21:05] or maybe I'm crazy. We spent time nailing this down so it would be right [21:05] but that's very illogical... [21:05] all that said, it looks wrong :-) [21:06] elfy, any last words? :P [21:28] knome, if you wouldn't mind righting our poor logic, we accept MP's; but also patches or simply corrected text :-) Bug reports welcome too [21:28] or i can just push the new version in the branch and the tracker? :) [21:29] and half of the point is that i'm not exactly sure what normal->LTS should do; my logic says "keep normal", but who knows... [21:35] knome, we like that option the best.. JFFI [21:36] ;) [21:40] balloons, done [21:41] <3 <3