| mwhudson | is it possible to tell launchpad which architecture to build a particular architecture: all package on? | 03:55 |
|---|---|---|
| sarnold | I've not seen that mentioned anywhere | 03:56 |
| sarnold | afaik it's fixed per release | 03:56 |
| wgrant | mwhudson, sarnold: XS-Build-Indep-Architecture works in LP and dak | 05:22 |
| wgrant | (it's a space-separated list of architecture tags, but usually just one) | 05:25 |
| slyon | hey LocutusOfBorg! I was just looking into the netplan.io autopkgtest failure on arm64, blocking network-manager 1.24 proposed migration... Seems like you got it working – was it a flaky test after all? | 06:41 |
| eoli3n_ | Is there any post which explains what's going on with snaps ? | 06:52 |
| eoli3n_ | i deploy my hosts with a kickstart install then i use ansible in "post" section. It fails installing chromium-browser because snap can't reach snap store. I think that's because i'm in a chroot | 06:53 |
| eoli3n_ | a package manager which needs a daemon ? is that docker who wrote that tool ? | 06:54 |
| eoli3n_ | https://bugs.launchpad.net/snappy/+bug/1609903 | 06:55 |
| ubottu | Launchpad bug 1609903 in Snappy "Enable the installation of snaps in a classic chroot" [High,Triaged] | 06:55 |
| eoli3n_ | is there a way to completly disable snapd without breaking anything ? | 06:58 |
| eoli3n_ | as i use ansible, i would be able to seperate snap tasks to not run in a chroot. But then, it is IN apt, what a choice ?? | 07:01 |
| eoli3n_ | should i test each of my 450 packages to know if it installs snap version or not ? | 07:01 |
| eoli3n_ | if i remove snapd package, then i try to install chromium for exemple, snapd comes as a dep to be able to install chromium | 07:09 |
| eoli3n_ | is that serious? | 07:09 |
| eoli3n_ | ubuntu isn't only used by geek teenagers who wants to rice gnome... | 07:10 |
| LocutusOfBorg | slyon, I just retried it... | 07:11 |
| LocutusOfBorg | flaky is the right wording I guess | 07:11 |
| andrewsh_ | eoli3n_: that’s because chromium is a snap package in Ubuntu | 07:12 |
| eoli3n_ | i noticed that andrewsh | 07:13 |
| andrewsh_ | if you want a non-snap package, you can install one from bionic-{updates,security} or Debian | 07:13 |
| andrewsh_ | I use the Debian one since the package from bionic doesn’t work with UberConference for some reason | 07:13 |
| eoli3n_ | andrewsh what i mean i that i need to be able to completly disable snap packages | 07:13 |
| andrewsh_ | I have snap completely disabled, yes | 07:14 |
| slyon | LocutusOfBorg, okay thanks! It was the first time it behaved this way.. It was re-tried many times before and now it suddenly works | 07:14 |
| eoli3n_ | because, if i add a new package to my playbook which requires snap, it will break my deployments | 07:14 |
| LocutusOfBorg | slyon, meh, I know, but some big migration happened before I retried it | 07:15 |
| LocutusOfBorg | we might meld the last two runs to see if something in toolchain made it pass | 07:15 |
| slyon | LocutusOfBorg, yes. I will investigate a little bit, trying to understand what was going wrong. | 07:16 |
| zyga | hello | 07:19 |
| zyga | my thinkpad x240 running groovy daily had some issues this morning - when I tried to log in via gdm I got kicked out of the session instantly | 07:20 |
| zyga | looking at the logs I noticed that X could not get /dev/dri/card0 | 07:20 |
| zyga | the file on my system is owned by root:video but my user account did not belong to the video group | 07:20 |
| zyga | I temporarily resolved this by adding my account to said group but I'm not sure this is right | 07:20 |
| zyga | it feels like something logind would ACL me into | 07:21 |
| zyga | seb128: ^ reported here | 07:21 |
| wgrant | Yeah, you'd normally get u:$USER:rw- on /dev/dri/card0 from logind, hm. | 07:21 |
| zyga | journal from this boot https://paste.ubuntu.com/p/TFJ6dqJzFK/ | 07:22 |
| zyga | line 4042 has: cze 05 08:58:04 x240 /usr/lib/gdm3/gdm-x-session[2867]: (EE) modeset(0): drmSetMaster failed: Permission denied | 07:23 |
| seb128 | zyga, thanks | 07:24 |
| zyga | hmm, my system has a 2nd user account for various tests | 07:25 |
| zyga | and it's possible that account has linger enabled | 07:25 |
| * zyga chceks | 07:25 | |
| zyga | yes | 07:26 |
| zyga | the 2nd account does have linger | 07:26 |
| eoli3n_ | https://bugs.launchpad.net/ubuntu/+source/snapd/+bug/1782873 | 07:51 |
| ubottu | Launchpad bug 1662552 in snapd (Ubuntu) "duplicate for #1782873 snaps don't work with NFS home" [Medium,Fix released] | 07:51 |
| eoli3n_ | who manage snapd integration ? | 07:54 |
| xnox | @pilot in | 08:01 |
| === udevbot changed the topic of #ubuntu-devel to: Archive: Open | 20.04 Released! | Devel of Ubuntu (not support) | Build failures: http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of Trusty-Focal | If you can't send messages here, authenticate to NickServ first | Patch Pilots: xnox | ||
| === diddledan0 is now known as diddledan | ||
| === ricab__ is now known as ricab | ||
| zyga | eoli3n_: hey | 09:10 |
| zyga | I should probably have a look | 09:10 |
| zyga | can you tell me how to replicate your setup? | 09:10 |
| zyga | I'm not a nfs user | 09:10 |
| eoli3n_ | Hi zyga | 09:11 |
| zyga | hi | 09:11 |
| eoli3n_ | is there a way to completly disable snaps ? | 09:12 |
| zyga | eoli3n_: remove snapd | 09:12 |
| zyga | eoli3n_: but if you tell me how to replicate the nfs setup I'm sure I can fix that bug | 09:12 |
| eoli3n_ | nop, if i install a package which requires snap install, it installs snapd as dep | 09:12 |
| eoli3n_ | my main problem is not NFS one, i didn't check yet if its still a problem, i will tell you for sure | 09:13 |
| zyga | what is the problem then? | 09:13 |
| eoli3n_ | apt purge snapd -y && apt install chromium-browser -y | 09:13 |
| zyga | chromium is not packaged as a deb anymore, nobody wanted to maintain it | 09:13 |
| zyga | it's a mess and pain to do | 09:14 |
| eoli3n_ | zyga, i use kickstart install, in %post i use ansible to install packages. As snaps requires snapd daemon, and i'm in a chroot it breaks my install | 09:14 |
| zyga | I see | 09:14 |
| zyga | perhaps file a bug on the chromium package to handle that case better (defer the install) | 09:15 |
| eoli3n_ | for now i only have the problem with chromium-browser, but what's fears me is that ubuntu will use more and more snaps packages | 09:15 |
| eoli3n_ | my concern isn't mainly about chromium-browser | 09:15 |
| eoli3n_ | it is about "how to massive deploy ubuntu with snap packages now integrated in apt" | 09:15 |
| zyga | I think there are ways to do that | 09:16 |
| zyga | but I'm not the best person | 09:16 |
| eoli3n_ | i need to use ansible in chroot, i always did and it works well | 09:16 |
| eoli3n_ | ubuntu just need to provide a way to disable snap packages | 09:16 |
| eoli3n_ | snaps are buggy still, force it will cause lots of problems | 09:17 |
| zyga | you can configure a seed to preinstall snaps | 09:17 |
| zyga | but I think you should not mix many problems at once | 09:18 |
| eoli3n_ | my problem is this : i use a playbook in a post kickstart chroot and that same playbook during host live : if i add a new package which is handled by snap, i will never see that and it will break my deployments | 09:19 |
| eoli3n_ | i don't want to test each packages i add | 09:19 |
| eoli3n_ | i'm really sad about ubuntu's choice, isn't docker a good light for "what you should not do" ? Why force snap to use a daemon, when flatpak shows you it is useless ? | 09:19 |
| zyga | perhaps someone else can help you, I can only help you with specific snapd topics | 09:20 |
| zyga | eoli3n_: I think you are wrong but I'm not sure you want to discuss it | 09:20 |
| zyga | you are unhappy about it and that's not something I can help with | 09:20 |
| eoli3n_ | yes you're right, lets don't talk about that point | 09:20 |
| zyga | snapd did not have a daemon, inititally it was just snappy, it was a huge mistake that took a year to fix | 09:21 |
| zyga | and there's a lot of really good reasons why | 09:21 |
| zyga | there are some ways to configure a system to install snaps on first boot (seeding) | 09:21 |
| eoli3n_ | ok ok, but you will face a lots of deployments problem as mine | 09:21 |
| zyga | we use it in ubuntu | 09:21 |
| eoli3n_ | hmm | 09:21 |
| zyga | perhaps, stuff will adjust though | 09:22 |
| eoli3n_ | but my concern isnt about "how to preseed a snap package" | 09:22 |
| zyga | we will help where we can | 09:22 |
| zyga | I know, you want things to work | 09:22 |
| zyga | I'm not familiar with ansible so I cannot really help you with that part | 09:22 |
| eoli3n_ | yes, for now i just need to find a way to "don't break my deployments" when adding a random package | 09:22 |
| zyga | perhaps someone else here could | 09:22 |
| eoli3n_ | ansible isn't the problem, consider it as a "apt install package" in a chroot | 09:23 |
| eoli3n_ | lets recenter | 09:23 |
| zyga | I think the package responsible for conversion should be fixed to detect this | 09:23 |
| zyga | and instead drop some hooks/triggers that go off when the system really boots | 09:23 |
| eoli3n_ | making snap installations offucated by apt is a bad choice | 09:23 |
| eoli3n_ | i will open an issue, can you tell me on which package i should open it ? | 09:24 |
| eoli3n_ | snapd ? | 09:24 |
| zyga | I think the way to make progress is to file a bug on the package to report that it misbehaves in a chroot | 09:25 |
| zyga | and carry on the discussion there | 09:25 |
| eoli3n_ | so chromium-browser ? | 09:25 |
| zyga | I think so | 09:26 |
| eoli3n_ | lets do this, thanks zyga | 09:27 |
| xnox | eoli3n_: for daemons we have policy-rc.d stuff. Which allows to download install, but not start, daemons. And it works in chroots. | 09:28 |
| xnox | eoli3n_: I guess for chroot based stuff we should have similar i.e. "download and add snap to seed.yaml / pending transaction" but do not try to start snapd or mount the snap. | 09:29 |
| xnox | eoli3n_: instead of installing Chromium package, you may want to call snap prepare-image to seed Chromium => that can be done in a chroot without snapd Daemon running. | 09:30 |
| eoli3n_ | xnox i don't get all you said | 10:13 |
| eoli3n_ | my main concern is "how to install a package without problem in chroot" | 10:13 |
| eoli3n_ | as apt wrap snaps and debs, i don't want the need to "do something to make it work" | 10:13 |
| eoli3n_ | i just want it to work | 10:13 |
| eoli3n_ | chromium is identified here, but my concern is about all existent packages. | 10:14 |
| xnox | eoli3n_: you may not realise, but installation of majority of packages in chroots already has detection code in maintainer scripts to not do things that break. | 10:15 |
| xnox | eo | 10:15 |
| eoli3n_ | oh ok, so that's a chromium-package issue | 10:15 |
| eoli3n_ | i didn't open issue yet | 10:15 |
| eoli3n_ | coffee time :) | 10:15 |
| xnox | eoli3n_: i.e. we make a lot of work, on every package, one by one, to ensure they install fine in chroot, lxd, container, vm, baremetal, WSL1, WSL2, over the network, on various filesystems, kernels, selinux confimenet, apparmor, etc. | 10:16 |
| xnox | eoli3n_: i am happy that for you it is perceived as "apt installs always work" =) | 10:16 |
| xnox | in general, and that we are doing things so good, that it is expected as normal =) | 10:17 |
| xnox | eoli3n_: but yeah, it feels like a bug against chromium-browser (or whatever the package is). To somehow do something sensible. | 10:17 |
| eoli3n_ | but shouldn't it be snap process which needs to catch chroot env ? | 10:18 |
| xnox | eoli3n_: i think installing lxd package should work in chroot. And i think that install creates stub scripts, that if invoked later, offer to install snap and run it, if snapd is running. | 10:19 |
| eoli3n_ | xnox: i didn't know about the work you talked about : so yes, great work :) | 10:19 |
| xnox | eoli3n_: right, that would be ideal if "snap install" would realise that it's in a chroot, that snapd is not running, and somehow "schedule to install this thing later, when snapd is operational" | 10:19 |
| xnox | eoli3n_: but that sounds like a separate bug/issue type of thing. | 10:20 |
| eoli3n_ | i will open it in chromium-browser package and see where it goes | 10:20 |
| xnox | (well feature) | 10:21 |
| oSoMoN | eoli3n, ping me with that bug once you've filed it, I'll be looking at it | 10:25 |
| eoli3n_ | nice, thanks to all of you ;) i'm kickstarting to paste error | 10:27 |
| eoli3n_ | oSoMoN https://bugs.launchpad.net/ubuntu/+source/chromium-browser/+bug/1882232 | 10:49 |
| ubottu | Launchpad bug 1882232 in chromium-browser (Ubuntu) "install chromium-browser in a chroot fails" [Undecided,New] | 10:49 |
| eoli3n_ | xnox: do you really want that i run in noninteractive ? | 12:35 |
| eoli3n_ | i'm on it , i will paste logs | 12:37 |
| dupondje | https://bugs.launchpad.net/ubuntu/eoan/+source/wireguard/+bug/1870293 | 12:47 |
| ubottu | Launchpad bug 1870293 in wireguard-linux-compat (Ubuntu Focal) "Resync wireguard/wireguard-linux-compat with development" [Undecided,Fix committed] | 12:47 |
| dupondje | Forgot to push it to -updates on Focal? | 12:47 |
| ahasenack | hi, focal has golang-go 1.14, but defaults to 1.13 via golang-defaults. Is there an "official" way to have 1.14 be the default, or should I create the symlinks that golang-defaults has manually? | 12:49 |
| ahasenack | I would like to use 1.14 in a package build, and I can't change the symlinks in the system in that build (not root), nor should I | 12:50 |
| ahasenack | I was hoping for some environment variables, like java | 12:50 |
| ahasenack | I can of course update PATH | 12:50 |
| xnox | eoli3n_: what you are experiencing, and reporting, does not sound like a bug. | 12:57 |
| eoli3n_ | why that ? | 12:58 |
| eoli3n_ | its normal that i can't install a package in a chroot ? | 12:58 |
| xnox | eoli3n_: many package installations pop-up debconf prompts upon installation, or upgrades, and require user input or acknoledgement by default. | 12:58 |
| eoli3n_ | xnox please read my answer | 12:58 |
| xnox | eoli3n_: this is not a failure to install. as apt didn't fail to install anything. | 12:58 |
| eoli3n_ | that's not a debconf issue | 12:58 |
| eoli3n_ | anyway, i'm running right now it in noninteractive mode | 12:59 |
| eoli3n_ | i'm waiting for timeout to break to be able to paste | 12:59 |
| xnox | eoli3n_: "-y" is not enough | 12:59 |
| eoli3n_ | i understood that | 13:00 |
| eoli3n_ | xnox... | 13:00 |
| eoli3n_ | please wait a min | 13:00 |
| xnox | eoli3n_: one has to either preseed debconf answer, or ask ansible to run in non-interactive mode | 13:00 |
| eoli3n_ | ansible runs defaulty in non-interactive mode | 13:00 |
| eoli3n_ | anyway as said, i'm facing that issue with DEBIAN_FRONTEND=noninteractive | 13:00 |
| eoli3n_ | if i'm reporting it, that's because there is a problem | 13:01 |
| xnox | eoli3n_: if install does not complete even with DEBIAN_FRONTEND=noninteractive => then there are bugs in maintainer scripts yes. | 13:02 |
| LocutusOfBorg | tjaalton, hello you there? | 13:02 |
| LocutusOfBorg | https://bugs.launchpad.net/ubuntu/+source/mesa/+bug/1876882 | 13:02 |
| ubottu | Launchpad bug 1876882 in libclc (Ubuntu Bionic) "Backport packages for 18.04.5 HWE stack" [Undecided,New] | 13:02 |
| LocutusOfBorg | looks like vbox is broken by new mesa? | 13:02 |
| xnox | eoli3n_: i'm sorry, but the bug report as was written was not clear at all. | 13:02 |
| eoli3n_ | the debconf pop up is clear | 13:02 |
| eoli3n_ | but i will add what you want | 13:03 |
| xnox | eoli3n_: the fact that there is debconf pop-up is valid and ok. Unless it is impossible to bypass them, when trying to do so. | 13:03 |
| xnox | which is what you seem to be reporting | 13:03 |
| eoli3n_ | did you read the pop up ? | 13:03 |
| eoli3n_ | "fails" means recturn code is 1 | 13:04 |
| eoli3n_ | anyway | 13:04 |
| eoli3n_ | sorry knox, but your answer has no sense | 13:04 |
| eoli3n_ | "-y" is not enough to install non-interactive under ansible. | 13:04 |
| eoli3n_ | ansible doesn't use -y or something | 13:04 |
| eoli3n_ | it has his own module | 13:04 |
| eoli3n_ | and it wraps a real noninteractive install | 13:05 |
| xnox | hang on | 13:08 |
| xnox | eoli3n_: the code in the template expects that dialog to have 3 options / answers | 13:08 |
| xnox | Retry Abort Skip | 13:08 |
| xnox | yet in your screenshot it's just "OK" | 13:08 |
| xnox | eoli3n_: which release / package version is this? | 13:09 |
| eoli3n_ | xnox i understand that you need to be sure that i don't say something wrong. But it would help if you help me to make you understand what's going wrong | 13:11 |
| eoli3n_ | the msg is clear | 13:11 |
| eoli3n_ | the problem is clear to | 13:11 |
| eoli3n_ | snapd is not running in chroot | 13:12 |
| eoli3n_ | and it make the package install fails | 13:12 |
| eoli3n_ | which release / package of what ? | 13:12 |
| eoli3n_ | i use netboot | 13:12 |
| eoli3n_ | so latest | 13:12 |
| eoli3n_ | which package ? apt updated so latest too | 13:13 |
| eoli3n_ | the problem is "Unable to contact the store, trying every minute for the next 30 minutes" | 13:14 |
| eoli3n_ | in debconf OR in text logs | 13:14 |
| eoli3n_ | you seems saying that i do something wrong: please reproduce then you will see :) | 13:15 |
| xnox | eoli3n_: and you expect to be able to use the chromium-browser in the chroot? or do you plan to boot that choot in a VM / bare mchine later | 13:15 |
| xnox | eoli3n_: let me attach my screenshot | 13:15 |
| xnox | eoli3n_: because it is different to yours | 13:15 |
| xnox | eoli3n_: which release of ubuntu are you using? | 13:15 |
| eoli3n_ | xnox i manage 800 ubuntu nodes for a university, so here i'm trying to migrate my deployment method | 13:16 |
| eoli3n_ | i install my whole system with kickstart then ansible in chroot | 13:16 |
| eoli3n_ | but why is that important | 13:16 |
| eoli3n_ | please try docker commands i gave you | 13:16 |
| eoli3n_ | and you will face that same issue | 13:16 |
| eoli3n_ | xnox: latest netboot | 13:17 |
| eoli3n_ | so 20.04 | 13:18 |
| * ogra thought latest netboot uses a livefs ... how does a chroot come into play here | 13:18 | |
| eoli3n_ | you seriously guys don't get what i do here ? aieaieaie ok lets explain more | 13:18 |
| eoli3n_ | im a sysadmin in a university | 13:18 |
| eoli3n_ | i deploy 800 ubuntu nodes on physical hosts for courses | 13:19 |
| eoli3n_ | i automated the whole process with kickstart | 13:19 |
| xnox | eoli3n_: https://launchpadlibrarian.net/482972565/debconf-prompt-groovy.png | 13:19 |
| eoli3n_ | kickstart deploys my system, then in %post section of kickstart i use ansible, to configure the whole system | 13:19 |
| eoli3n_ | then it reboots and the host is ok and working | 13:19 |
| xnox | eoli3n_: so i am confused how come I am offered "retry abort skip ok" and you are only offered "ok" | 13:20 |
| xnox | eoli3n_: and you will need to preseed the answer to Skip with debconf preseed (aka automatic ahead-of-install answers to package questions) | 13:20 |
| eoli3n_ | xnox ansible wrap it | 13:21 |
| ogra | xnox, different debconf priority settings ? | 13:21 |
| eoli3n_ | it does what you say | 13:21 |
| xnox | eoli3n_: what ansible settings do you have for the debconf module in your playbook? https://docs.ansible.com/ansible/latest/modules/debconf_module.html | 13:21 |
| eoli3n_ | i don't use the debconf module | 13:22 |
| xnox | eoli3n_: you must, if you want ansible to answer the debconf questions for you | 13:22 |
| eoli3n_ | didn't even know it exists | 13:22 |
| tjaalton | LocutusOfBorg: yes? | 13:22 |
| eoli3n_ | xnox that's strange : i use that method since 5 years | 13:22 |
| eoli3n_ | never had to use debconf module | 13:23 |
| xnox | eoli3n_: https://paste.ubuntu.com/p/2N78xNW2m8/ | 13:23 |
| xnox | eoli3n_: it means your installations are extremely trivial. | 13:23 |
| xnox | (despite however complicated they are) | 13:23 |
| xnox | from dpkg / debconf point of view at least | 13:23 |
| eoli3n_ | so your solution, is to wait for that long timeout | 13:23 |
| eoli3n_ | then tell to install package later ? | 13:24 |
| xnox | eoli3n_: no | 13:24 |
| xnox | eoli3n_: if you use debconf, ansible will preseed that answer for you, and the timeout will not even be attempted. | 13:24 |
| xnox | eoli3n_: it will instantly skip and carry on. | 13:24 |
| eoli3n_ | so package will not be deployed ? | 13:24 |
| xnox | eoli3n_: this is what debconf preseeding is for, to answer things ahead of time. | 13:24 |
| tjaalton | LocutusOfBorg: where's the full log | 13:24 |
| xnox | eoli3n_: the package will be installed. | 13:24 |
| xnox | eoli3n_: the snap will not be, until later when users boot the system | 13:25 |
| eoli3n_ | sorry, i think the part i don't get is "what snap is trying to do with that timeout ?" | 13:25 |
| xnox | eoli3n_: also why are you installing the package at all? | 13:25 |
| xnox | eoli3n_: this has nothing to do with snap | 13:25 |
| eoli3n_ | because an OS is to run packages ?? | 13:25 |
| xnox | eoli3n_: Ubuntu runs on snaps and debs. | 13:25 |
| eoli3n_ | i don't get what you mean | 13:25 |
| eoli3n_ | so ? | 13:25 |
| eoli3n_ | xnox: eoli3n_: also why are you installing the package at all? | 13:26 |
| xnox | eoli3n_: i.e. chromium-browser deb is there for convenience. When preparing Ubuntu Studio images with chromium-browser, instead of isntalling it as a deb, we preseed it as a snap using seeded snaps functionality. | 13:26 |
| eoli3n_ | any link about that ? | 13:26 |
| eoli3n_ | that snaps start to bored me | 13:27 |
| xnox | it's easier to preseed snaps on first boot, directly from the store without trying to use debconf + apt modules of ansible to take three detours to install a snap. | 13:27 |
| eoli3n_ | what a mess | 13:27 |
| eoli3n_ | seriously | 13:27 |
| eoli3n_ | KISS powa | 13:27 |
| xnox | eoli3n_: https://docs.ansible.com/ansible/latest/modules/snap_module.html | 13:28 |
| xnox | - name: Install browser | 13:28 |
| eoli3n_ | xnox my concern is : how to know a package IS a snap if its wrapped by apt | 13:28 |
| xnox | snap: | 13:28 |
| xnox | name: | 13:28 |
| xnox | - chromium-browser | 13:28 |
| xnox | and done | 13:28 |
| eoli3n_ | so at each new package i install : i need to verify that that's a deb or snap package | 13:29 |
| eoli3n_ | that's your solution ? | 13:29 |
| eoli3n_ | so why integrated snaps in apt if its better to keep them separated ? | 13:29 |
| xnox | eoli3n_: majority of things only exist in one form. (i.e. only a deb, or only a snap). And we try to minimise number of things that are available as wrappers. | 13:29 |
| xnox | eoli3n_: because it is confusing. | 13:29 |
| eoli3n_ | ubuntu: "cool, we integrated snaps in apt" me: "there's an issue" ubuntu: "separate your installs then" | 13:29 |
| xnox | eoli3n_: however, we must provide them for upgrades and to transition people off a deb into a snap. | 13:30 |
| === hggdh is now known as hggdh-msft | ||
| xnox | eoli3n_: chromium-browser is not intended to be installed on fresh machines. It is only intended to support upgrades and conversions. | 13:30 |
| xnox | eoli3n_: so far in the archive we have 3 packages that are wrappers around snaps, for upgrades: maas lxd chromium-browser | 13:31 |
| eoli3n_ | ok, but so what about my issue ? https://bugs.launchpad.net/ubuntu/+source/chromium-browser/+bug/1882232/comments/6 | 13:31 |
| ubottu | Launchpad bug 1882232 in chromium-browser (Ubuntu) "install chromium-browser in a chroot fails" [Undecided,Incomplete] | 13:31 |
| xnox | eoli3n_: interractively we do guide people to the right tool. | 13:31 |
| xnox | eoli3n_: and grpahically snap-store UX is clear, it is a one stop place that installs either snaps or debs. | 13:31 |
| xnox | eoli3n_: your issue is valid. The experience is bad, frustrating and confusing. | 13:31 |
| eoli3n_ | i can confirm that | 13:32 |
| xnox | eoli3n_: and requires a lot of details explaining as to what is happening, why, and how to do things better. | 13:32 |
| eoli3n_ | i'm trying to debconf solution | 13:32 |
| xnox | eoli3n_: and we must improve it. given that there is too much information out there that says "apt install chromium-browser" | 13:32 |
| eoli3n_ | xnox so write a post about that somewher | 13:32 |
| xnox | eoli3n_: let me rephrase/rewrite that bug report and keep it open. | 13:32 |
| eoli3n_ | ok thanks for your help | 13:33 |
| eoli3n_ | add the deconf workaround please | 13:33 |
| eoli3n_ | and "chromium-browser-l10n" should be installed as package yes ? | 13:34 |
| eoli3n_ | but it has chromium-browser as dep in apt | 13:34 |
| xnox | chromium-browser-l10n => is empty | 13:34 |
| xnox | Description: Transitional package - chromium-browser-l10n -> chromium snap | 13:34 |
| xnox | This is a transitional dummy package. It can safely be removed. | 13:34 |
| xnox | . | 13:34 |
| xnox | chromium-browser-l10n is now replaced by the chromium snap. | 13:34 |
| eoli3n_ | ok thanks | 13:35 |
| xnox | eoli3n_: on fresh installs you only want to "snap install chromium-browser" on first boot | 13:35 |
| xnox | (with ansible module, or with a cron job or etc) | 13:35 |
| xnox | eoli3n_: or if you must be able to complete it offline there is a way to populate /target/var/lib/snapd directory with files there to do it for | 13:35 |
| eoli3n_ | but will i face the same issue in chroot if i use snap module ? | 13:35 |
| LocutusOfBorg | tjaalton, https://launchpad.net/~costamagnagianfranco/+archive/ubuntu/locutusofborg-ppa/+build/19411432 https://launchpad.net/~costamagnagianfranco/+archive/ubuntu/locutusofborg-ppa/+build/19411433 | 13:35 |
| xnox | eoli3n_: let me finish typing update to the bug report. | 13:36 |
| eoli3n_ | ok | 13:36 |
| LocutusOfBorg | here some detail https://bugzilla.redhat.com/show_bug.cgi?id=1765930 | 13:36 |
| ubottu | bugzilla.redhat.com bug 1765930 in mesa "recent mesa / libglvnd changes break mutter + mutter328 builds" [Unspecified,Closed: currentrelease] | 13:36 |
| eoli3n_ | i'm trying the snap module in chroot | 13:36 |
| xnox | eoli3n_: not sure. I do not quite know how that ansible snap module works. Ideally, it would be smart and either does "snap install" or "snap prepare-image" (for cached offline install) | 13:36 |
| xnox | eoli3n_: if it does the first thing, then it will fail too =/ | 13:36 |
| eoli3n_ | lets try, just to know | 13:37 |
| xnox | eoli3n_: what is your full installation? do you already have anything that stores things on disk (in that choot) and is executed on first boot? | 13:39 |
| eoli3n_ | i don't get what you mean? | 13:39 |
| eoli3n_ | my playbook in 500 tasks | 13:39 |
| eoli3n_ | so yes | 13:39 |
| xnox | but all of them run in a choot, and later you reboot into that chroot? | 13:40 |
| xnox | after the install is done / all the tasks are done / machine deployed? | 13:40 |
| eoli3n_ | i never reboot in the chroot, as the chroot is just the post install script | 13:40 |
| eoli3n_ | let me paste something | 13:40 |
| xnox | eoli3n_: but like it's not possible to run desktop or browser from a chroot. without X, etc. | 13:41 |
| xnox | eoli3n_: so what is the point of it? (i kind of want to know the full picture of what you are doing) | 13:41 |
| eoli3n_ | xnox i don't get what you think, but seriously that's strange | 13:41 |
| eoli3n_ | do you know how kickstart works ? | 13:41 |
| xnox | no | 13:41 |
| eoli3n_ | that's the main problem | 13:41 |
| ogra | its just a debconf frontend in ubuntu ... | 13:41 |
| eoli3n_ | i pxe netboot ubuntu installer | 13:42 |
| eoli3n_ | i provide a "answer file" | 13:42 |
| eoli3n_ | which install in non interactive way ubuntu | 13:42 |
| xnox | i know that people boot kernel+initrd, run installer to provision the hard-drive which during install is a mounted chroot, and at the end the said machine is rebooted, from disk, and people get a graphical desktop. | 13:42 |
| eoli3n_ | that's what i do yes | 13:42 |
| eoli3n_ | expect that at the end of install | 13:42 |
| eoli3n_ | i use %post kickstart to run ansible in chrooted env | 13:43 |
| eoli3n_ | then i reboot | 13:43 |
| eoli3n_ | and i get a full deployed host as you said | 13:43 |
| xnox | eoli3n_: hence my question, when that machine boots for the first time for the end user is there something running on first boot only? i.e. connect / enroll to ldap, configure printer, generate ssh keys, etc? | 13:43 |
| xnox | or are they all "golden" and pristine? | 13:43 |
| eoli3n_ | the same playbook runs at each reboot | 13:43 |
| eoli3n_ | so yes | 13:43 |
| xnox | do they enroll into master puppet / ansible / etc to be remote controleld administered. | 13:44 |
| eoli3n_ | but it has nothing to do because it freshly ran | 13:44 |
| eoli3n_ | no nothing of that | 13:44 |
| eoli3n_ | when the host first boot, it is fully configured | 13:44 |
| xnox | eoli3n_: but when it re-runs on the reboot, it still operates the same way from chroot, or does it run from rootfs? | 13:44 |
| xnox | right, i see. | 13:44 |
| xnox | thanks. | 13:44 |
| eoli3n_ | xnox: read this -> http://ix.io/2olM | 13:45 |
| eoli3n_ | everything after "%post" key runs in freshly installed chroot | 13:46 |
| eoli3n_ | that complete file is passed to pxeboot env | 13:46 |
| eoli3n_ | so it automates the whole process | 13:46 |
| eoli3n_ | then it reboots, and everything is working at first boot | 13:46 |
| oSoMoN | I am looking at r-cran-gwidgetstcltk test failures, and I think I have a fix which I'm going to submit to Debian in a moment | 13:56 |
| xnox | eoli3n_: by the way, all the things that start with "preseed " word are debconf questions answered. | 14:02 |
| xnox | eoli3n_: i.e. preseed chromium-browser/blah-blahs select Skip => should be usable there. but let me read it in full | 14:02 |
| === smoser1 is now known as smoser | ||
| xnox | eoli3n_: you are doing a lot of interesting things there! it is very neat! | 14:03 |
| tjaalton | LocutusOfBorg: why does it import both.. | 14:07 |
| tjaalton | LocutusOfBorg: you could sync the internal header to match mesa | 14:07 |
| tjaalton | LocutusOfBorg: like this https://gitlab.freedesktop.org/mesa/mesa/-/commit/3dd299c3d5b88114894ec30d1fac85fba688201f | 14:09 |
| xnox | cpaelzer: vorlon: i remember somebody somewhere asked before how to reproduce i386-amd64 autopkgtest builders | 14:21 |
| xnox | and i'm failing to find instructions again | 14:21 |
| xnox | was it you? do you remember where it was posted on how to do it? | 14:21 |
| xnox | https://code.launchpad.net/~ubuntu-release/autopkgtest-cloud/+git/autopkgtest-cloud/+ref/i386-as-amd64-cross ? | 14:22 |
| Eickmeyer | xnox: Just for reference, we don't seed chromium in Ubuntu Studio in any way, but I hope the example still got across. :) | 15:05 |
| LocutusOfBorg | tjaalton, will try to fix thanks | 15:15 |
| eoli3n_ | thx xnox | 15:35 |
| xnox | Eickmeyer: yeah, i know. hence using studio images is no problem. but trying to do, what studio iso is doing, is hard. because it is impossible to discover that one must do "snap ack --remote model:generic..... >model ; snap prepare-image model --snap core18 --snap snapd --snap chromium-browser" | 15:41 |
| eoli3n_ | lets drink a bear and continue monday morning :) have a nice we | 15:41 |
| eoli3n_ | a beer | 15:41 |
| eoli3n_ | not a bear :D | 15:41 |
| Eickmeyer | xnox: Oh, yeah, totally. | 15:42 |
| Eickmeyer | xnox: I also need to bug #ubuntu-release to figure out what is going on with our Focal images, it seems to be installing two kernels at once and livecd-rootfs is freaking out and failing. | 15:42 |
| === heliocastro is now known as helio|afk | ||
| === ben_r_ is now known as ben_r | ||
| sarnold | wgrant: oh wow, cool, I've never seen XS-Build-Indep-Architecture before :) thanks | 17:37 |
| === Guest52001 is now known as Enlik | ||
| xnox | it's quite new | 18:08 |
| xnox | wgrant: is it fully supported in lp now? | 18:09 |
| sarnold | after reporting it to #debian-til, I got back a suggestion that it's not really in debian; they said something along the lines of, dak doesn't manage the buildds.. | 18:19 |
| xnox | sarnold: yeah, that's what i thought it was more or less work in progress to support that | 18:36 |
| cpaelzer | xnox: I was the one who asked i386 autopkgtest | 19:05 |
| cpaelzer | xnox: it comes down to an amd64 image + extra arch | 19:05 |
| cpaelzer | xnox: I put the info on https://wiki.ubuntu.com/i386 which is AFAIK the one place i386 info is supposed to be | 19:05 |
| xnox | cpaelzer: thanks | 19:06 |
| xnox | i hate fftw3 and somehow i fail at autopkgtest | 19:38 |
| xnox | i should give up on it | 19:38 |
| xnox | @pilot out | 19:38 |
| === udevbot changed the topic of #ubuntu-devel to: Archive: Open | 20.04 Released! | Devel of Ubuntu (not support) | Build failures: http://qa.ubuntuwire.com/ftbfs/ | #ubuntu for support and discussion of Trusty-Focal | If you can't send messages here, authenticate to NickServ first | Patch Pilots: | ||
| cjwatson | xnox: It's been fully supported in LP since 2015. | 19:49 |
| wgrant | xnox: since 2014 or so I think | 22:31 |
| === hggdh-msft is now known as hggdh | ||
| xnox | ooooh nice | 22:51 |
| * xnox ponders if i can use it for cd-boot-images then! | 22:52 | |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!
