pitti | Good morning | 05:11 |
---|---|---|
didrocks | good morning | 05:55 |
pitti | bonjour didrocks | 06:01 |
didrocks | hey pitti, how are you? | 06:03 |
pitti | didrocks: quite well, thanks! juggling systemd, langpacks, and urgent apport features :) | 06:07 |
didrocks | good luck! | 06:07 |
pitti | et toi ? | 06:07 |
didrocks | I'm fine, thanks | 06:08 |
larsu | morning :) | 06:24 |
pitti | hey larsu, wie gehts? | 07:11 |
seb128 | hey pitti & desktopers | 07:12 |
seb128 | pitti, wie gehts? | 07:12 |
* pitti ^5s seb128, gut danke! und dir? | 07:12 | |
seb128 | pitti, larsu is off for the morning, he's have spanish classes this week (and shifted hours) | 07:12 |
* seb128 ^5s pitti back, good, danke! | 07:12 | |
pitti | seb128: ah, I missed his "good morning" earlier | 07:12 |
seb128 | hum | 07:54 |
seb128 | so that iso doesn't boot in kvm :-/ | 07:54 |
seb128 | "systemd: Started udev Coldplug all Devices." | 07:54 |
seb128 | it just sits there | 07:54 |
seb128 | 13s in the boot | 07:55 |
Laney | hello! | 08:06 |
didrocks | morning Laney | 08:07 |
seb128 | hey Laney | 08:08 |
seb128 | pitti, is boot blocking on "systemd: Started udev Coldplug all Devices." symptomatic of some known class of system misconfiguration/bug? | 08:11 |
Laney | hey didrocks & seb128 | 08:13 |
seb128 | Laney, how is u.k today? | 08:13 |
Laney | cloudy, rain later | 08:15 |
Laney | and nl? | 08:15 |
seb128 | sunny today! | 08:18 |
seb128 | and not too windy, which is nice | 08:18 |
pitti | seb128: not really -- that sounds like a kernel driver issue usually, but hard to tell without more logs | 08:22 |
pitti | seb128: drop "quiet splash" and boot with "debug systemd.debug-shell"? | 08:22 |
seb128 | pitti, k | 08:23 |
seb128 | pitti, it hangs on http://imagebin.ca/v/25WyUYshyvpy ... no way to go to vt8 or something | 09:18 |
pitti | seb128: hm, is that with a new kernel or something? since when did that happened? | 09:28 |
Laney | bet a thousand euros it's the personal image | 09:30 |
Laney | :) | 09:30 |
seb128 | Laney, you won! | 09:30 |
seb128 | pitti, it's a candidate of snappy personnal image I'm trying to build in kvm | 09:31 |
seb128 | it's the first image build that worked so can't compare to "previous versions/since" | 09:31 |
pitti | seb128: ah, ok; and no VT switching? seems that something is breaking stuff really hard then :/ could you try moving /etc/sysctl.d/ away for testing if it's one of the knobs there which trigger a kernel fault or so? | 09:32 |
seb128 | pitti, how do you I move things away without a command line? can I do that from grub? | 09:47 |
seb128 | booting with upstart it hits cloud-init expection errors and seem to bug on that | 09:47 |
pitti | seb128: you can't boot rescue either? | 09:47 |
seb128 | some mkdir failing | 09:47 |
pitti | seb128: you could boot with init=/bin/bash | 09:47 |
pitti | then mount -o remount,rw / | 09:48 |
pitti | and h4ck stuff there | 09:48 |
seb128 | let me try recovery | 09:48 |
pitti | I can help you in a bit, I'm just rather swamped ATM | 09:48 |
seb128 | don't worry, thanks for the hints | 09:49 |
pitti | and Laney's "fix udisks" club is becoming a real threat too | 09:49 |
pitti | seb128: in the mean time, feel free to scp it to people.c.c.? | 09:49 |
seb128 | pitti, it's a 10G image, going to take a while on my dsl line, going to do that over lunch | 09:49 |
pitti | seb128: convert it to qcow2? | 09:50 |
pitti | seb128: please don't upload a raw image :) | 09:50 |
seb128 | k | 09:50 |
* seb128 googles | 09:50 | |
seb128 | qemu-img convert -f raw -O qcow2 | 09:51 |
pitti | seb128: qemu-img convert -O qcow2 xx.raw xx.img | 09:51 |
pitti | seb128: you can then also use -snapshot | 09:51 |
pitti | when running the VM, to avoid modifying it | 09:51 |
* didrocks has some easy-to-use scripts for this if interested | 09:52 | |
seb128 | didrocks, converting? | 09:52 |
seb128 | shoot! :-) | 09:52 |
didrocks | seb128: not converting, but starting with/without snapshot, using hw accel… | 09:52 |
* Laney hugs pitti | 09:53 | |
Laney | feel free to punt that if you have some hints about where to look | 09:53 |
didrocks | seb128: http://people.canonical.com/~didrocks/vms/vmcommon | 09:53 |
seb128 | didrocks, I'm interested! | 09:53 |
didrocks | seb128: http://people.canonical.com/~didrocks/vms/ rather | 09:53 |
seb128 | didrocks, 'ci | 09:53 |
pitti | fun how everyone has their own :) | 09:53 |
didrocks | seb128: make them all executable, but vmcommon | 09:53 |
seb128 | k | 09:54 |
pitti | http://people.canonical.com/~pitti/scripts/vm and http://people.canonical.com/~pitti/scripts/vmbr here | 09:54 |
didrocks | Usage: /home/didrocks/bin/vm [--master] <image> [vm options...] | 09:54 |
pitti | so that I can do "vm /my/ubuntu.img -snapshot" | 09:54 |
pitti | and vmbr is with using the lxcbr0 bridge for "proper" networking | 09:54 |
Laney | you crazy people | 09:54 |
didrocks | interesting, mine are -snapshot by default, unless you specify --master to modify base image | 09:54 |
Laney | I just load it up in virt-manager :P | 09:54 |
didrocks | (as I'm using snapshot way more) | 09:55 |
pitti | didrocks: yeah, I tend to do that too, good idea | 09:55 |
didrocks | the vm-net is inspired by the commands you gave me some times ago, seems similar to vmbr :) | 09:55 |
pitti | yeah, "vm" just searches a free port and uses that for ssh redirection, vmbr does the bridge thing | 09:55 |
didrocks | Laney: remember that you are the one liking tiling wm :) | 09:57 |
larsu | hi pitti :) | 10:57 |
larsu | man it's nice out | 10:57 |
ogra_ | in berlin too ? | 11:00 |
pitti | c'était gris en Bavière d'hier :( | 11:04 |
larsu | ogra_: ya. Perfect. | 11:11 |
ogra_ | kassel too :) | 11:12 |
larsu | nice :) | 11:12 |
pitti | Laney: udisks is an actual regression; investigating now | 11:24 |
pitti | perhaps related to newer kernel/util-linux/udev, as udisks2 itself hasn't changed in a while | 11:24 |
=== MacSlow is now known as MacSlow|lunch | ||
=== alan_g is now known as alan_g|lunch | ||
=== alan_g|lunch is now known as alan_g | ||
=== MacSlow|lunch is now known as MacSlow | ||
=== rickspencer3_ is now known as rickspencer3 | ||
seb128 | pitti, ok, so I managed to get a debug console, tons of jobs fail to start :-/ | 14:28 |
seb128 | journalctl has a lot of "failed dependencies", is there a way to tell which one? | 14:28 |
seb128 | also trying to start e.g getty@tty7 returns a job for ... canceled" | 14:29 |
larsu | it is configured to only allow 6 gettys | 14:29 |
larsu | this is on touch? | 14:29 |
pitti | seb128: can you pastebin a journalctl output? | 14:30 |
* pitti needs to run for some errands, back in 20 | 14:30 | |
seb128 | larsu, ignore getty, I can list any other job in the same situation, rc-local.service if you prefer | 14:31 |
seb128 | pitti, not easy, I'm in a debug console of a kvm without much access to outside | 14:31 |
pitti | seb128: systemctl --failed | 14:31 |
seb128 | larsu, and no, it's not touch, it's snappy personnal try | 14:31 |
seb128 | pitti, that lists 38 failed jobs | 14:31 |
seb128 | but mostly mount ones | 14:32 |
seb128 | which I know why they fail, but that shouldn't block the system to start, should it? | 14:32 |
larsu | depends on which ones fail I guess | 14:32 |
seb128 | larsu, /writable touch rw overwrites | 14:33 |
seb128 | not system partitions | 14:33 |
larsu | hm? | 14:33 |
seb128 | hum? | 14:33 |
larsu | huum? | 14:33 |
seb128 | what do you "hm?" about? | 14:33 |
larsu | not sure what you mean by "/writable touch rw overwrites" | 14:33 |
seb128 | the fstab includes bindmounts for touch | 14:33 |
seb128 | which don't work on snappy | 14:33 |
seb128 | but I don't really care if those locales stay ro | 14:34 |
seb128 | so those jobs failing is fine to me (if that doesn't block other things) | 14:34 |
larsu | ah, that makes sense | 14:34 |
larsu | haven't dug that deep into systemd yet | 14:34 |
larsu | but probably it's configured to not boot when mounts fail | 14:34 |
seb128 | is there a "retry all failing jobs?" ;-) | 14:35 |
larsu | I don't think so, no :) | 14:35 |
larsu | there's probably some mount target that you can activate though | 14:36 |
* larsu looks | 14:36 | |
larsu | seb128: what does `systemctl list-units --type=targets --all` say? | 14:38 |
larsu | especially for the local-fs.target | 14:38 |
seb128 | "unknown unit type or load state 'targets'" | 14:39 |
larsu | sorry, it's --type=target | 14:39 |
seb128 | all is not-found | 14:40 |
larsu | not good :/ | 14:40 |
seb128 | syslog as well | 14:40 |
seb128 | everything else is loaded | 14:40 |
seb128 | local-fs is loaded inactive dead | 14:40 |
seb128 | which I guess is not good | 14:40 |
larsu | indeed | 14:41 |
larsu | should be active :) | 14:41 |
larsu | you can try reactivating it, but I'm not sure why that would work now | 14:41 |
seb128 | yeah, a dependency job failed | 14:42 |
seb128 | but it doesn't say which one | 14:42 |
larsu | not even in the jounral? | 14:42 |
larsu | journalctl -f in a separate terminal is really handy btw | 14:43 |
seb128 | yeah, just too much content in there | 14:44 |
seb128 | like it lists 50 jobs that fail | 14:44 |
seb128 | k, that's ridiculous, I guess I need to clear out the fstab/writable issue first and see what remains | 14:46 |
pitti | seb128: if fstab mounts fail that aren't marked with "nofail", they will fail local-fs.target and with it pretty much everything else | 14:46 |
seb128 | pitti, yeah, I guess it's the issue there | 14:46 |
seb128 | pitti, can I edit fstab and tell "retry all the jobs"? ;-) | 14:47 |
seb128 | or retry start local-fs ? | 14:47 |
pitti | seb128: (untested) "systemctl default" | 14:47 |
pitti | seb128: but might be easier to just reboot :) | 14:47 |
pitti | seb128: systemctl status local-fs.target doesn't say which dependency failed? it shold | 14:48 |
seb128 | pitti, hum, are changed to qcow2 images stored/persistent? | 14:51 |
pitti | seb128: normally yes, unless you start qemu with -snapshot | 14:51 |
pitti | usually better to start with -snapshot, as you can hack away without destroying the image | 14:52 |
seb128 | k | 14:52 |
seb128 | pitti, http://picpaste.com/qemu-7fiqV7Bv.png | 14:52 |
didrocks | or if you use my script… :) | 14:53 |
* seb128 tries reboot | 14:53 | |
pitti | seb128: ok; then "systemctl --failed --type=mount" should list the failed ones | 14:54 |
seb128 | pitti, yeah, there is like 38 of those | 14:54 |
seb128 | ok, boots now to what looks like a display server fail, progress! | 14:55 |
seb128 | didrocks, do you remember why you didn't include http://bazaar.launchpad.net/~ubuntu-core-dev/livecd-rootfs/trunk/view/head:/live-build/ubuntu-core/hooks/20-extra-files.chroot in the ubuntu-desktop-next dir? | 14:57 |
seb128 | that seems the issue | 14:57 |
seb128 | the /writable dir is missing so all the mount job fail | 14:57 |
seb128 | jobs* | 14:57 |
pitti | seb128: ah, nice! | 14:58 |
pitti | ln -sf /proc/mounts /etc/mtab | 14:58 |
pitti | oh for $DEITY's sake, can we just completely eradicate /etc/mtab? | 14:59 |
pitti | it seems snappy is the one thing that doesn't need to be concerned about 30 year old tools for disks/mounts administration? | 14:59 |
didrocks | seb128: once sec, looking (sorry, finishing up opening a BTS issue on licensecheck) | 14:59 |
seb128 | didrocks, thanks | 14:59 |
seb128 | pitti, yeah, I just want the "mkdir -p /writable" | 15:00 |
seb128 | mvo_, ogra_, http://bazaar.launchpad.net/~ubuntu-core-dev/livecd-rootfs/trunk/view/head:/live-build/ubuntu-core/hooks/20-extra-files.chroot ... see pitti's comment about mtab | 15:00 |
didrocks | seb128: maybe a mistake. I remember to have deleted one file on purpose, but I don't think it was that one | 15:00 |
pitti | mvo_, ogra_: specifically, did you just add /etc/mtab "because we always did", or does something on snappy actually need it? if the latter, can we fix that? | 15:01 |
ogra_ | seb128, yes, and ? i'm fully with him | 15:01 |
ogra_ | pitti, no, i had to add it else mountall tries to set it up | 15:01 |
pitti | (I'm volunteering for fixing it, FTR) | 15:01 |
ogra_ | we discussed that, remember | 15:01 |
pitti | ogra_: aah, ok | 15:01 |
seb128 | ogra_, so it's needed? | 15:01 |
pitti | ogra_: so that was just to quiesce mountall | 15:01 |
ogra_ | it needs to be a sym,link | 15:01 |
pitti | but that doesn't affect snappy, just touuch? | 15:01 |
ogra_ | we used to do that on the s-i server when re-packing | 15:01 |
ogra_ | does systemd have anything handling mtab ? | 15:02 |
ogra_ | i think you said no | 15:02 |
pitti | ogra_: our package has a "fixup" job to replace a file with a symlink | 15:02 |
pitti | upstream, no | 15:02 |
ogra_ | then snappy wont need it | 15:02 |
pitti | the last bit of util-linux was fix in 2.25 I believe (but we only turned on that option in wily) | 15:02 |
pitti | ogra_: ah, I just see that we do need to adjust /lib/systemd/debian-fixup | 15:03 |
pitti | we don't need /lib/systemd/system/debian-fixup.service in snappy at all, so if we are looking for some 0.01 s optimization somewhere.. :) | 15:04 |
seb128 | ogra_, pitti, so should I keep it or delete it? | 15:04 |
pitti | seb128: delete it, and I'll fix debian-fixup.service | 15:05 |
ogra_ | pitti, btw, i discussed mtab with slangasek when i added that ... (we are trying to get rid of it since at least 5 years) and i think he said it would probably time to re-visit that ... iirc there are still some debs that rely on it and iots format he said | 15:05 |
pitti | it'll fail on a r/o image until then, but that's harmless | 15:05 |
seb128 | pitti, k, what happens without your fix? | 15:05 |
ogra_ | seb128, /etc isnt writable ... | 15:06 |
pitti | seb128: debian-fixup.service will likely fail as it can't create the /etc/mtab link | 15:06 |
ogra_ | so whatever would try to create it would fail | 15:06 |
seb128 | k, but that doesn't block the boot? | 15:06 |
pitti | seb128: no, should be fine; it's just a wants | 15:06 |
seb128 | great | 15:06 |
ochosi | Laney: pressed the power button on my new xps13 for the first time one minute ago | 15:07 |
seb128 | ogra_, mvo_, do you want it out of ubuntu-core as well? | 15:07 |
seb128 | ochosi, oh, you received it? great :-) | 15:07 |
ochosi | yep! :) | 15:07 |
ogra_ | seb128, i dont reallyy care, just leave it in touch | 15:07 |
ochosi | first time i see ubuntu OEM | 15:07 |
Laney | ochosi: nice, good luck ;-) | 15:07 |
ochosi | Laney: well i've already downloaded a 15.04 iso, but i still wanted to see what the ootb experience was | 15:08 |
ogra_ | the point of this hook is to make it possible to use plain cdimage rootfses without them being re-packaged on the s-i server | 15:08 |
ogra_ | i thinnk it will make mountall fall over or at least unhappy on the phone | 15:08 |
ochosi | Laney: is it just Dell or do you ship chrome by default these days? | 15:11 |
Laney | haha of course not | 15:11 |
seb128 | ochosi, just oem images | 15:11 |
Laney | that's them | 15:11 |
ochosi | thought so, just wanted to be sure | 15:11 |
seb128 | we have discussions every now and then about chromium by default but it never managed to make it default | 15:12 |
ochosi | yeah, basically it'd mean replacing tb as well, no? | 15:12 |
ochosi | at least to get the full benefit of dropping XUL | 15:12 |
seb128 | no | 15:12 |
seb128 | yeah, well that's not the main driver motivation to change | 15:13 |
seb128 | the people who want chromium default just think that it has more users/is more popular | 15:13 |
ogra_ | well, it will soon have less ads :) | 15:14 |
ogra_ | (given the new firefox ad model) | 15:14 |
ochosi | Laney: whoa, who made BIOS look like Windows 95?? | 15:16 |
seb128 | pitti, ok, next problem, the vm boots to what looks like x failing to start, but it loops on that and no way to go to a vt or anything, any clue how to debug? | 15:33 |
seb128 | recovery does the same | 15:33 |
seb128 | tried to boot with init=/bin/bash but that hangs boot and doesn't give me a shell | 15:34 |
pitti | seb128: can you boot with console=ttyS0, and with qemu option "-serial stdout" | 15:34 |
pitti | sorry, "-serial stdio" | 15:34 |
pitti | seb128: then you will get a serial console and the boot messages there | 15:34 |
pitti | and don't need to rely on the internal VTs | 15:35 |
seb128 | pitti, great, danke | 15:36 |
seb128 | arg | 15:38 |
seb128 | don't ctrl-C on that std :p | 15:38 |
seb128 | it stops the qemu | 15:38 |
pitti | seb128: yeah :/ | 15:40 |
pitti | seb128: do you get some useful output and a getty there? | 15:40 |
seb128 | pitti, yes, basically xorg fails to start as expected | 15:41 |
seb128 | it just do it in loops for ever | 15:41 |
seb128 | and in recovery mode as well? (that feels wrong) | 15:41 |
pitti | so lightdm keeps restarting? or the autologin does? | 15:41 |
pitti | yes, recovery shouldn't start lightdm and default.target and all that | 15:41 |
seb128 | lightdm keeps restarting as far as I can tell | 15:42 |
ogra_ | seb128, hmm you dont create the firmware and modules mountpoints ? | 15:43 |
ogra_ | (why ?) | 15:43 |
seb128 | ogra_, is that needed and why? | 15:43 |
seb128 | because I don't understand why that's here | 15:43 |
seb128 | and it's undocumented | 15:43 |
ogra_ | to give your system access to modules and firmware files :) | 15:43 |
seb128 | so I try without and if it's needed I'm going to know why and document it | 15:43 |
ogra_ | that is where the initrd mounts them | 15:43 |
ogra_ | s/mounts/bind mounts/ | 15:43 |
seb128 | so it's needed? | 15:44 |
ogra_ | if you plan to use any modules or firmware ... yes | 15:44 |
ogra_ | well, if you always only use system-image re-packed rootfses it will currently still work because the s-i server re-packs the whole rootfs just to add these three dirs | 15:44 |
ogra_ | but we are planning to drop that code from there | 15:45 |
=== alan_g is now known as alan_g|AFK | ||
ogra_ | this hook was in preparation of the droppage | 15:45 |
seb128 | oh ok | 15:45 |
ogra_ | so if you want it future proof, better create these two dirs too | 15:45 |
seb128 | I was going to say, livecd-rootfs copy things to those dirs | 15:46 |
seb128 | k | 15:46 |
seb128 | noted for the next upload, thanks | 15:46 |
ogra_ | no, it doesnt ... they are just nodes for mountpoints | 15:46 |
ogra_ | (firmware and modules come from the device tarball, separate from the rootfs) | 15:46 |
seb128 | ogra_, well, on the booted image those dirs exist and have content from the kernel image deb | 15:47 |
ogra_ | http://bazaar.launchpad.net/~ubuntu-system-image/ubuntu-system-image/server/view/head:/lib/systemimage/generators.py#L525 | 15:47 |
ogra_ | thats the code in system-image the hook replaces | 15:47 |
ogra_ | (and which we will drop) | 15:47 |
seb128 | pitti, yeah, according to the journal lightdm does "Failed with result 'exit-code'" and it keeps retrying | 15:48 |
seb128 | ogra_, k, thanks | 15:48 |
mvo_ | seb128: could you do me a favour and reject my accidental golang-gettext from vivid-proposed. that was a mistake | 15:51 |
seb128 | mvo_, done | 15:52 |
mvo_ | ta | 15:52 |
seb128 | yw | 15:53 |
ochosi | Laney: oh wow, installing 15.04 cleanly doesn't seem like much fun... (uefi and no wifi) | 15:56 |
ochosi | and sry, i'll stop spamming you with this now | 15:56 |
Laney | ochosi: I just got bcmwl on there manually | 15:56 |
Laney | and it is installed in efi mode | 15:57 |
ogra_ | 15.04 isnt even usable on the old XPS13 | 15:57 |
* ogra_ will actually do a reinstall of trusty on the weekend | 15:57 | |
ochosi | Laney: switching to legacy left me with: floppy, cd/dvd, usb and mini-sd as boot options... | 15:58 |
=== alan_g|AFK is now known as alan_g | ||
seb128 | Laney, can you retry desktop-next iso build?. | 16:58 |
Laney | try using your powers? | 16:58 |
seb128 | Laney, where is the button? | 17:00 |
seb128 | I don't see any control on https://launchpad.net/~ubuntu-cdimage/+livefs/ubuntu/wily/ubuntu-desktop-next | 17:00 |
ogra_ | ssh nusakan :) | 17:01 |
Laney | wrong and wrong | 17:01 |
Laney | go to http://iso.qa.ubuntu.com and log in | 17:01 |
Laney | then go to "Wily daily" | 17:01 |
Laney | tick the boxes next to the things you want to rebuild | 17:01 |
ogra_ | pfft | 17:01 |
seb128 | shrug | 17:02 |
seb128 | login fails "The name seb128 is already taken." | 17:02 |
larsu | lol | 17:02 |
Laney | haha | 17:02 |
Laney | did it use sso? | 17:02 |
larsu | other seb :D | 17:02 |
seb128 | "Have you forgotten your password?" points to http://iso.qa.ubuntu.com/user/password | 17:02 |
seb128 | which is "The resource you tried to access doesn't exist. This can happen because of an invalid link or because of a bug." | 17:02 |
seb128 | great | 17:02 |
seb128 | quality website | 17:02 |
Laney | be charitable please | 17:03 |
seb128 | sorry ;-) | 17:03 |
Laney | probably go ask stgraber, that is his product | 17:03 |
seb128 | can you click for me then? ;-) | 17:03 |
Laney | fine, if you go ping him so that I don't have to keep doing this forever | 17:04 |
Laney | probably good to see if doing it via this button works first anyway ... | 17:04 |
* Laney always did it over ssh before | 17:04 | |
seb128 | Laney, done on -devel | 17:04 |
Laney | thx | 17:05 |
seb128 | thank for pointing me to the website | 17:05 |
Laney | didn't work anyway | 17:07 |
Laney | #FAIL | 17:07 |
seb128 | :-/ | 17:09 |
seb128 | stgraber is looking at my issue | 17:09 |
seb128 | it's better I'm older than the sso | 17:09 |
* seb128 dinosaur | 17:09 | |
=== alan_g is now known as alan_g|EOD | ||
Laney | :) | 17:32 |
Laney | I retried it the old way for now | 17:32 |
seb128 | Laney, thanks | 17:34 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!