=== sem2peie- is now known as sem2peie === sem2peie- is now known as sem2peie [08:30] doko: I see that you have been fixing symbols due to LTO. presumably that mean these fails are not to be added to the lto-disabled list? [08:30] ^ moved from wrong channel [08:54] RikMills: yes, fixing these for main directly [08:55] doko: and universe shoudl be fixed or excluded? [08:55] specifically symbols [08:56] e.g. https://launchpad.net/ubuntu/+source/kpmcore/20.12.3-2 [08:59] RikMills: universe packages are added in the lto-disabled list. If not, add it there, or fix it. For the main packages, I added (optional=lto) to the symbols files [09:00] doko: well, I will probably fix this one, as I want to make sure it is in the beta. then it can be added to the list and resynced with debian sometime [09:00] thanks! [09:06] RikMills: ok. if you're able to apply these symbol changes in Debian, please do that. https://wiki.debian.org/ToolChain/LTO [09:07] problem is, that Debian doesn't yet have a dpkg with the optimize feature [09:10] Hi, with automatic-ubiquity, is there a way to disable splash ? [09:10] i mean, the GUI, to just have tty logs [09:10] doko: yeah, so I doubt I will be able to sell the qt-kde people the change at this stage of freezes [09:12] oh well. doing things ahead of them while they are in freeze is not exactly a new thing for me ;) [09:49] tsimonq2 Hi https://wiki.ubuntu.com/UbiquityAutomation [09:49] i think that this needs to be improved [09:50] for exemple, i don't get what is ubiquity/summary, nor how to use it [09:50] ubiquity ubiquity/summary boolean true ? string yes ? select yes ? [09:50] what is "summary page" [09:51] you should add an exemple with comments for all keys [09:51] how to query list of all possible values ? [09:51] nothing says that preseed/late_command is ignored, and it seems to be replaced by ubiquity/success_command [13:39] xnox: Aha, TIL about group.master and base-passwd: I was looking in base-files for something similar. Thanks! (Thanks cjwatson too!) [13:39] xnox i have some questions if you have some time [13:47] - is there a way to disable installer UI and keep tty and logs ? [13:48] - shouldn't the installer get host's hostname from DHCP by default ? [13:49] - as first question, do you know a way to print output of success_command ? [13:50] with kickstart, I used "exec < /dev/tty6 > /dev/tty6 2> /dev/tty6" then chvt 6 before running ansible [13:51] but then with ubiquity, i get /dev/tty6 not found [13:51] i think that the chroot doesnt mount /dev [13:58] eoli3n: there is in-target helper to do more mounts i think. [13:58] eoli3n: "print output of success_command"where to? =) [13:58] on the scree [13:58] eoli3n: normally syslog & journal are connected and captured. [13:58] screen :) [13:59] eoli3n: so for example, anything in journal/logger will end up in the journal of the live system, and also copied into /var/log/installer/* [13:59] i need technicians to easily see what's going wrong [13:59] i found logs in /var/log/syslog [13:59] eoli3n: well, you booted graphically =) there is ubiquity window to see running log, but it's not expanded by default. [13:59] how to not boot graphically then ? [14:00] eoli3n: why would not? =) i think you are more after setting more kernel cmdline options no? [14:00] thoses logs are just /var/log/installer/debug [14:00] why would not what ? i didn't get what you meant [14:00] i.e. set systemd.journal to go to console. and like chvt to tty4 then your technicians will be staring on the console with journal things running there. [14:01] eoli3n: booting into graphical ubiquity interface is best. Because when it crashes, it gives you popups to report bugs; see tracebacks; recover the install; and one has access to open gnome-terminal / web-browser / etc. [14:02] ok, but is there a way to boot "not graphically" ? [14:02] not bringing up graphical interface will make it _harder_ to debug, not easier. [14:02] i really need one to NOT access anything [14:02] eoli3n: not really, cause ubiquity.service runs on a graphical tty [14:02] that's sad but ok [14:03] then i don't get your answer about "print output of success_command" [14:03] you can only like switch to a different tty, and keep things "hidden", but like anybody can do ctrl-alt-f2 to get back to it. [14:03] eoli3n: you say you want to print output of success_command => but you never said where to =) [14:03] my suggestion is to run success_command under logger -t => such that any output ends up in syslog/journal. [14:03] i said "on the screen" [14:04] keyboard is disabled [14:04] i need those logs to be print on the screen [14:04] normally automatic installs are non-interactive and nobody is watching the screen.... [14:05] eoli3n: so you don't like switching tty and printing journal there? [14:05] and have success command log to journal, which is displayed in a different tty? [14:05] eoli3n: note that one should be able to click on the expand window to see the running installer logs. [14:05] anything from journal is visible there, but requires a user click. [14:06] xnox as said, where installer is running, keyboard is disabled [14:06] maybe in automatic install, that pane should be always expanded [14:06] eoli3n: i understand that keyboard is disabled, but your success command can execute a binary which changes vt _without any keyboard presses_ =) [14:07] yep that's what i tried too [14:07] didn't work? [14:07] you also need kernel parameters to mirror journal to active console. [14:07] 14:51 eoli3n: but then with ubiquity, i get /dev/tty6 not found [14:07] xnox where can i find documentation about this ? [14:08] about "kernel parameters to mirror journal to active console" [14:08] and about "in-target helper to do more mounts i think." [14:08] its really really hard to find any good documentation about all of these [14:08] eoli3n: man systemd-journald => KERNEL COMMAND LINE => systemd.journald.forward_to_console=, systemd.journald.forward_to_wall= [14:09] ok thanks [14:10] eoli3n: in-target is binary shipped by d-i & ubiquity.... I don't think it is documented, but you can just read what it does. [14:10] or like run-it interactively and see if you like what it did..... or not. [14:10] i.e. $ sudo in-target bash [14:10] oh ok thanks for that tip [14:10] after you finish the interactive install..... but did not reboot yet. [14:11] didn't know it was a binary [14:12] last thing is about hostname by dhcp [14:14] eoli3n: i think ubuqity derives hostname from autogenerated based on chasis name. Can you preseed it? [14:15] statically i can't [14:15] or i don't understand what you mean [14:15] i.e. in your preseed, when you do like partman/early_command => you should have dhcp hostname acquired at that point [14:15] i do something like that, but i don't understand why i need to do it manually [14:15] so you would want to do soemthing like `debconf-set-selection netcfg/hostname string `hostname -fqd` [14:15] host $(/usr/sbin/ip a show dev $(/usr/sbin/ip route show default | awk 'sub(/.*dev /,"",$0) {print $1; exit;}') | awk '/inet / {sub(/\/.*/,"",$2); print $2}') | awk 'sub(/\..*/,"",$NF) { print $NF }' > /etc/hostname [14:16] i don't know that command, lets check [14:16] "-fqd" [14:16] eoli3n: because default is not based on the network name, but is based on chaiss/machine name. It is more userfriendly to say "Dell-XPS13-ubuntu" then "ubuntu.192.168.1" cause most networks don't give out sensible hostnames. [14:17] eoli3n: i'm not sure what i've typed is correct => this is just me going off by memory. [14:17] eoli3n: doing that ip stuff is a bit odd..... ie. in success_command you could do just: `hostname` > /target/etc/hostname should do it. [14:18] but hostname command alone doesnt give the right hostname... [14:18] should it ? [14:19] i get "ubuntu" as default hostname [14:19] eoli3n: hmmmmm. and dhcp lease deffo had hostname from dhcp? interesting. [14:19] eoli3n: what about output from hostnamectl => does that have dynamic hostnmae from dhcp? [14:19] lets check [14:19] rebooting [14:20] strangly i also had some issues about using fqdn in url= [14:20] then i switch to ip [14:21] from url=http://domain.fr/preseed.cfg to url=http://1.1.1.1/preseed.cfg [14:22] Laney: I've got a SRU on my plate for update-notifier for focal - but it looks like there is some existing work of yours there that hasn't propogated. Do you know what needs to happen next? [14:22] dbungert: thanks I need to verify that, forgot, will do [14:23] :) [14:23] eoli3n: it would be interesting to know the details of the lease you got. [14:23] from what I saw it was a bystander on a update-manager test failure, but update-manager has since propogated. Might be as simple as a retest. [14:24] xnox https://x0.at/wj2.png [14:24] eoli3n: i.e. stuff in network-manager and/or in systemd (in like /run/systemd/netif ) [14:24] eoli3n: and what's your kernel cmdline? [14:24] eoli3n: this looks odd [14:25] xnox https://x0.at/8qL.txt [14:25] eoli3n: that looks borked to me =) [14:25] what means borked [14:25] eoli3n: and what's in your /run/systemd/netif ? if anything? [14:25] eoli3n: borked => broken, as in i would have expected you to gain dhcp hostname. [14:27] ok, /run/systemd/netif/leases/2 contains HOSTNAME=right_hostname [14:28] eoli3n: fun. cause then hostnamectl should have it too...... [14:28] ok. who broke the builders> [14:28] and hostname..... [14:28] ? [14:28] eoli3n: please open a bug report? live system booted with ip=dhcp has the right lease, yet hostname not set at all. [14:29] https://x0.at/fdx.png [14:29] xnox i had the same issue with autoinstall with ubuntu-server [14:29] eoli3n: like ubuntu-bug systemd [14:29] eoli3n: given that it uses the same bits for networked boot, i can see why it's the same issue with server too. The networking stack is the same. [14:29] lets open the bug [14:30] $ ubuntu-bug systemd [14:32] yep [14:32] thaks [14:32] thanks [14:41] xnox just saw that i already opened one without any answer [14:41] reopened this : https://bugs.launchpad.net/ubuntu/+source/systemd/+bug/1920933 [14:41] Launchpad bug 1920933 in systemd (Ubuntu) "live desktop system booted with ip=dhcp has the right lease, hostname not set at all" [Undecided,New] [14:41] at the end i linked that one : https://bugs.launchpad.net/subiquity/+bug/1905932 [14:41] Launchpad bug 1905932 in subiquity "autoinstall: dynamic hostname from DHCP for the installer" [Undecided,New] [14:52] xnox again thanks for your precious help [14:56] dbungert: done, now someone needs to release it === _hc is now known as Guest21860 === ijohnson is now known as ijohnson|lunch === Guest21860 is now known as _hc [18:26] bdmurray: just a heads up: we uploaded livecd-rootfs 2.664.19 for focal-proposed yesterday and is awaiting approval. (For https://bugs.launchpad.net/ubuntu/+source/livecd-rootfs/+bug/1920043) Any expedition is greatly appreciated. [18:26] Launchpad bug 1920043 in livecd-rootfs (Ubuntu) "Latest Focal Minimal disk images from livecd-rootfs do not boot" [Undecided,New] [19:01] who would be the right person to ping to get some trivial changes to the sponsoring report reviewed? [19:01] https://code.launchpad.net/ubuntu-sponsoring/+activereviews [19:03] (basically moving the json to a subdir so they can be listed without knowing the names and adding to extra information to those) [19:03] also who is maintaining the machine hosting the report? [19:03] I would be interested to know what Ubuntu version is in used there and if the sponsoring code could be upgraded to python3 [19:04] I'm wanting to provide a patch, I'm just unsure which python version is available [19:07] pitti, hey! do you plan to release a new python-dbusmock version? I'm asking because of https://ftp.acc.umu.se/pub/GNOME/sources/gnome-bluetooth/3.34/gnome-bluetooth-3.34.5.news (the new integration test are relying on new mocks) === ijohnson|lunch is now known as ijohnson [19:30] seb128: let me see if its on the server I think its on [19:31] bdmurray, thanks [19:31] bdmurray, also can you help with the trivial code reviews? ;) [19:33] seb128: its on cranberry which is precise but python3 is available [19:34] bdmurray, ok, thanks [19:34] seb128: I could do a trivial review this week [19:35] bdmurray, well if you find the time I would appreciate, https://code.launchpad.net/ubuntu-sponsoring/+activereviews . One is just moving the location of the .jsons and the other adding some extra info to those, I want to use those for the KPI [20:43] doko, thanks for the evolution-data-server fix, do you plan to send it to Debian? the package was in sync and we would like to keep it this way