[00:46] can anyone tell me where debian-installer hides the information about which is the default keyboard layout for a language? [00:47] xnox: thanks for your reply. I ended up having to go deal with work stuff. i'll take a look. though a few of the systems i'm stuck working with are pre-systemd. Thanks for the pointers though. === _hc is now known as Guest78499 === Guest78499 is now known as _hc [09:33] santa_: filed https://bugs.launchpad.net/ubuntu/+source/glibc/+bug/1915250 [09:33] Launchpad bug 1915250 in glibc (Ubuntu) "buildd file owner/group for shared libraries " [High,Confirmed] [09:33] rbalint: ^^^ [09:33] I have no clue yet why this is only seen for a few packages [09:38] mwhudson: d-i keyboard-configuration, and there's also something relevant in languagechooser - the selection of default locale for a language which GunnarHj basically looks after for us these days [11:08] doko: thanks a lot [11:28] teward: /etc/fstab should work with pre systemd too. [11:28] teward: i.e. upstart & mountall should honor options to /run & /dev too [11:29] mwhudson: not debian installer! a random inline patch in ubiquity. [11:29] mwhudson: but i think that's just language, not keyboard layout (aka swiss case) [11:29] mwhudson: https://git.launchpad.net/ubiquity/tree/d-i/patches/localechooser-post-base-installer.patch not sure if this is helpful [14:29] tseliot: https://bugs.launchpad.net/ubuntu/+source/nvidia-graphics-drivers-460/+bug/1915145 i attached the patch there, as on github i can't find 460 packaging for hirsute. [14:29] Launchpad bug 1915145 in nvidia-graphics-drivers-460 (Ubuntu Hirsute) "add modalias for testing" [Undecided,New] [14:29] tseliot: i did upload that into my ppa, and in qemu, with the smbios tweaks, i can now run `ubuntu-drivers list` and it senses that nvidia-460 is needed. [14:30] tseliot: it would help us a lot with nvidia installer testing; and will allow automating nvidia installer tests in the lab. [14:30] tseliot: can you add that to all supported releases, next time you are doing uploads? [14:30] i am happy to help verify those srus. [15:03] Is any core dev available to restart the failed pyfuse3 autopkgtests that are blocking python3-defaults? There is a new version of pyfuse3 in hirsute that should resolve the failures in the tests. [15:28] coreycb: Does the openstack team have any plans to merge python-ceilometerclient? https://merges.ubuntu.com/main.html?query=ceilo&showProposed=true&showMergeNeeded=true [15:30] bdmurray: we have a todo to catch up with some merges, I'll add that to our card [15:31] coreycb: cool, thanks [15:31] JawnSmith: I'll have a look at that [15:37] xnox, looking [15:39] xnox, why don't you copy my code from ubuntu-drivers-common instead (test_ubuntu_drivers.py)? https://pastebin.ubuntu.com/p/tQfmzt7FVs/ [15:42] xnox, also this (from the same file) https://pastebin.ubuntu.com/p/Mvy8XmT6QP/ [15:54] seb128, xnox: are the libzip test failures triggered by the new openssl? [15:58] tseliot: that's bad, that needs to happen inside the booted Live ISO [15:59] tseliot: i want to set those smbios things on a qemu-kvm and run auto-preseeded ubiquity iso test with an unmodified installer .iso from cdimage.ubuntu.com for point release testing and boom it magically trying to trigger nvidia & oem logic of the install. [15:59] tseliot: i.e. i am trying to force this externally without umockdev installed in the live environment (because it isn't). [15:59] tseliot: does that make sense? [16:00] cause i'm faking certified platofrms with qemu (i.e. normally set vendor Lenovo, version blah-blah, etc.) [16:00] to trigger the oem meta installs, which depend on the sku. [16:01] xnox, well, I assume you are creating a fake dmi device to test things [16:08] tseliot: dmi is not fake, it's something that stock qemu/libvirt supports since forever. [16:08] tseliot: there are no fake pci devices, hence i can't create a qemu device which has pci id / modalias matching existing nvidia cards. [16:09] tseliot: if there is a way to create a virtual random qemu device that matches those modaliases that will be nice.... but also i think kernel will try to load modules on them and wreck havoc. [16:09] tseliot: hence i want a modalias that is in the metadata but not in the kernel module. [16:09] xnox, if I add that modalias, won't autoinstalling nvidia break things in qemu? [16:10] tseliot: no. [16:10] tseliot: by default qemu provides zero smbios. [16:11] xnox, oh, ok. Which driver series are you going to test in the installer? The latest? What about the other drivers? [16:11] tseliot: and it is a user provided setting for smbios (aka dmidecode data) to set that to UBUNTUNVIDIATEST which we all agree is the thing to trigger nvidia module installation. Nothing by default sets that in smbios when launching qemu vms; but our custom libvirt. [16:11] tseliot: so i was thinking that maybe we need two modalias. [16:12] tseliot: UBUNTUNVIDIATEST and UBUNTUNVIDIA460TEST => that way i can launch a vm that will be compatible with all nvidias, or just the 460 one. [16:12] tseliot: testing just the latest is a good start. [16:12] tseliot: cause like my laptop is compatible with 3 out of 4 nvidias that we ship. [16:12] hence i was thinking to set UBUNTUNVIDIATEST modalias on all nvidias that we ship. [16:13] tseliot: if you want us to test them all, we can do that too =) [16:13] but then we'd want nvidia specific version strings. and possibly universal one too to emulate real devices that are compatible with multiple nvidias. [16:16] xnox, the main thing is that those series will be migrated to newer series. Currently we have a legacy series (390), 450, and 460 (at least the ones on the desktop side of things) [16:17] tseliot: oh yeah, that thing. [16:17] tseliot: as if it makes more sense to generate something like dmi pvr UBUNTUNVIDIA$card-idTEST => where card-id is one of the ids supported by said driver. [16:18] to mimick the GeForce MX350 or whatever. [16:18] and some other card-id for whichever card still uses 340 [16:19] tseliot: i hope you can pick appropriate testing modalias that makes the most sense =) as long as it's a dmi pvr one that qemu/libvirt can fake [16:21] xnox, right, I will have to think about that [16:28] bdmurray: thanks! [16:31] could you possibly also restart python-limits? It previously failed with a bunch of packages from proposed, but it appears to succeed if the only proposed package is python3-defaults [17:27] in testing a potential SRU to fix a Perl related crash issue in NGINX, is there a way to trigger all Perl autopkgtests against a PPA version of Perl with the patch in it? Because if I remember right when Perl gets put into the queue it gets a massive insane number of autopkgtests that get triggered. [17:45] teward: This talks about how to run one package against a PPA https://wiki.ubuntu.com/ProposedMigration#Testing_against_a_PPA [17:47] I don't know how to get the list of what would be triggered [17:48] JawnSmith: python-limits submitted === ijohnson is now known as ijohnson|lunch [18:28] We have a thunderbird, tinyjsd and jsunit versioning question. [18:28] Both packages are gone in Groovy. [18:28] Does the thunderbird package need a versioned Breaks against them in Groovy? [18:29] We are doing that in Focal, because we're going to ship empty packages for them in an SRU in Focal, and the thunderbird SRU should push those up. [18:30] And is a versioned Breaks sufficient? [18:30] I'm not clear on how package removals interact with do-release-upgrade, autoremove, etc. [20:21] ItzSwirlz: yes please generate the other two. It means I don't have to write changelog messages in your name. [20:23] alright. Sorry for the confusion or whatnot === ijohnson|lunch is now known as ijohnson [20:54] Alright. pbuilder-dist is running it's process, i gtg so when im back at home ill upload the debdiffs [21:11] nevermind [21:12] rbasak: debdiffs are ready and have been uploaded to lp bug