/srv/irclogs.ubuntu.com/2015/03/30/#ubuntu-devel.txt

=== _salem is now known as salem_
=== salem_ is now known as _salem
=== gusnan is now known as Guest58714
janimo85125604:48
pittiGood morning05:24
tnkhanhdo people use switch-case nowadays?05:33
tnkhanhI always use if-else05:33
smbhallyn, thanks06:57
=== spineau is now known as spineau_afk
Laibschmicahg: do you have a minute?07:27
micahgLaibsch: sure07:28
Laibschre bug 1438025. I admit to not being 100% certain about the proposed change, either. But the link you posted makes me believe that it is correct even more.07:29
ubottubug 1438025 in apt-cacher-ng (Ubuntu) "please add security.ubuntu.com to the ubuntu mirrors" [Undecided,Invalid] https://launchpad.net/bugs/143802507:29
Laibschsecurity.ubuntu.com is only a subset of the packages available in the true mirror network, I believe.  The purpose is to allow for quicker distribution of security-related updates.07:30
LaibschAs such, anything available in security.u.c is a subset of the mirror pool.07:30
micahgLaibsch: I think the idea is that individuals shouldn't be mirroring security.ubuntu.com and that the updates are actually copied to the -updates pocket within a short amount of time07:30
LaibschCorrect?07:30
Laibschthe proposed change wouldn't have that effect07:31
Laibschquite the opposite07:31
Laibschit would lower the burden on security.u.c07:31
Laibschat least that is my understanding and experience of acng07:32
micahgisn't that list used for pre-caching07:32
Laibschnope, not at all07:32
Laibschit defines a list of equivalents07:32
Laibschso, if a package is already in the cache, then acng will not download it again if requested from a mirror07:34
Laibschlet's say user A requests package P version 1.1 from de.archive.ubuntu.com. That package is now cached.  It was a security update so it's available on security.ubuntu.com as well.  User B now requests the same package, but uses security.ubuntu.com for the request.  acng will then download the file over the network one more time (and store it separately.07:36
Laibschit will NOT treat the files as equivalents, unless security.ubuntu.com is listed as a "Ubuntu mirror" (even if only partial mirror).07:37
micahginteresting...I'll reopen and subscribe mdeslaur, he can help clarify this I believe as he's more familiar with acng, thanks for the followup07:40
LaibschThanks07:40
LaibschAgain, I'm certainly not 100% certain about the effects of it, either07:40
LaibschWe should wait for input from upstream Eduard as well07:40
Laibschprecisely, about what happens if one of the mirrors is only a partial mirror07:41
Laibschthanks, micahg. Have a nice day.07:41
micahgLaibsch: thanks, you too07:41
=== spineau_afk is now known as spineau
tjaaltonis it just me or can't the kvm instance graphics device be changed from the gui? it just complains that the RAM option is only available for QXL (which seems to be the default on vivid)07:54
tjaaltonlooks like virt-manager is a bit old too, could be fixed upstream so I'll just leave it at that08:03
zygahey08:04
* zyga still sees wlan0 being used instead of eth0 when both are active on vivid08:04
zygathis is my route output http://paste.ubuntu.com/10705612/08:05
zygaaww08:05
zygasill me08:05
zygathat was after I turned it off08:05
zygahttp://paste.ubuntu.com/10705615/08:05
zygathat's with wifi enabled but I think it's only true on fresh boot08:06
darkxstzyga, odd, I get the opposite, like 2 default routes when both wifi and eth are active08:06
darkxstthat breaks just as bad though08:06
zygadarkxst: that might be what I see too, this is after clicking on the indicator applet08:07
darkxstzyga, ubuntu GNOME doesnt have indicator applet08:08
zygadarkxst: well, it's network manager on the back too08:10
dholbachgood morning08:12
darkxstzyga, in my case only wifi is managed by network manager08:12
dz0nyzyga: remove all routes for vmnet08:13
zygadz0ny: how?08:13
dz0nyroute del :)08:14
zygadz0ny: I see this as a bug in one of the components of the stack, I updated from utopic a few weeks ago and that was not a problem there08:14
darkxstdon;t see why vmnet routes would cause an issue08:14
zygadz0ny: I can remove them but this should not need to be done each time08:14
zygadz0ny: and I think darkxst is right, traffic goes through wifi, not vmnet*08:15
dz0nyvmnet1 is treated as link local08:16
dz0nyi think you have some script that does that08:16
darkxstdz0ny, they come from vmware08:17
zygadz0ny: vmnet are from vmware workstation,08:17
dz0nyit is used to push all packets to vmware palyer08:17
zygadz0ny: though they look okay08:17
darkxstmay or may not be bridged depending on config08:17
zygathey are on different networks08:17
zygavmnet1 is host-only, vmnet8 is nat08:18
dz0nymhm08:18
LocutusOfBorg1good morning08:18
dz0nyi think its vmware issue08:18
zygadz0ny: how can it be a vmware issue where all the traffic goes through wlan0?08:19
darkxstdz0ny, i don't think its vmware issue, Ive seen the double default route issue on my laptop that doesnt even have vmware installed08:59
Riddellmvo: so any thoughts on the apt upgrade issue we are having?09:29
flexiondotorgcjwatson, Thanks for making the new release of grub-gfxpayload-lists10:25
mgedminwill there be a new gfxboot-theme-ubuntu release with updated launchpad translations, before 15.04?10:25
=== doko_ is now known as doko
Odd_Blokedhclient is no longer running in the background in the vivid (proposed) images I'm looking at; is this expected because of (network|system)d10:33
Odd_Bloke*?10:33
Odd_Bloke(It did run at boot; it has output in the journal)10:33
Odd_Blokepitti: ^?10:35
pittiOdd_Bloke: no, not expected; but systemd just calls /etc/init.d/networking start, so might be a more recent change in ifupdown or dhclient?10:37
Odd_BlokeAck; I'll have a poke around.10:38
pittiOdd_Bloke: I have dhclient -1 on current vivid cloud images10:38
pitti(for eth0)10:38
pittiand we don't use networkd, at least it has never been discussed and it'd be rather late trying to integrate it now10:39
flexiondotorgcjwatson, Is it possible to specific a particular version of a package for a given architecture in a seed?10:40
flexiondotorg*specify10:40
dz0ny  695 ?        Ssl    0:01 /usr/sbin/NetworkManager --no-daemon10:43
dz0ny  788 ?        S      0:00  \_ /sbin/dhclient -d -q -sf /usr/lib/NetworkManager/10:43
dz0ny  792 ?        S      0:01  \_ /usr/sbin/dnsmasq --no-resolv --keep-in-foregroun10:43
* dz0ny vivid10:43
pittiflexiondotorg: no, that wouldn't make much sense -- if the version matches the archive it's redundant, and if it doesn't match you couldn't build an image at all10:46
=== MacSlow is now known as MacSlow|lunch
flexiondotorgpitti, OK.10:58
flexiondotorgpitti, I only ask because Xorg is completely busted for PowerPC since the recent update. I wanted to try reverting in the image build.10:59
pittiflexiondotorg: we can't do that, as the previous version isn't even in the archive any more10:59
cjwatsonExperimentation like that is better done locally.11:04
flexiondotorgcjwatson, Yeah, we are doing. I was just try to find an easier for testers to help.11:11
flexiondotorgpitti, Have you seen anything like this? - https://bugs.launchpad.net/ubuntu/+source/mate-settings-daemon/+bug/143772211:11
ubottuLaunchpad bug 1437722 in mate-settings-daemon (Ubuntu) "ubuntu MATE 15.04: mate-settings-daemon problem with systemd" [Undecided,New]11:11
flexiondotorgpitti, I can't reproduce on i386, amd64 or armhf.11:12
flexiondotorgpitti, Is systemd-shim still required in 15.04?11:12
pittiflexiondotorg: no, I haven't seen that; we use upstart for the session, so that shouldn't change (and sudo systemctl org.mate.settingsd.service wouldn't make sense)11:13
pittiunless mate actually does use systemd for the session?11:14
pittiflexiondotorg: no, we don't need shim if you run with systemd11:14
dz0nysystemctl start org.mate.settingsd.service --user ?11:14
pittibut we still want to support booting with pstart11:14
dz0nyisn't that user service?11:14
pittiyes, it is for sure11:15
flexiondotorgpitti, OK. indicator-datetime Depends on systemd-shim11:16
flexiondotorgpitti, Which is why I asked if it was required because I couldn't figure out why it is installed.11:16
flexiondotorgpitti, Regarding systemd, MATE really only uses logind, if it is available.11:17
pittiflexiondotorg: uh, indicator-datetime still depends on systemd-services; that's gone since at least trusty11:17
flexiondotorgpitti, The systemd delay in startup seem to be attributed to audio in the forum dicussion.11:18
Odd_Blokepitti: It looks like systemd 219-5ubuntu1 is the cause of dhclient no longer running in the background; it's the only relevant change in the image in which we started seeing the problem: http://cloud-images.ubuntu.com/vivid/20150327.3/unpacked/vivid-server-cloudimg-amd64-20150327.2-to-20150327.3-manifest.diff11:26
pittiOdd_Bloke: ok; would you mind filing a bug with the details? it's running here with today's cloud image, so it's not entirely obvious to me what you mean11:27
Odd_Blokepitti: As in `pgrep dhclient` returns something/11:28
Odd_Bloke*?11:28
Odd_BlokeI am really struggling with question marks today.11:28
* pitti tosses a shift key to Odd_Bloke11:28
pittiyes, I have dhclient -1 running, as expected11:28
pittiOdd_Bloke: can I sees the image version in the installed systemd anywhere?11:29
pittiargh, and I can't write "system" any more for the life of me11:29
ogra_pitti, just use the phone kbd, it reliably turns it into systems11:29
Odd_BlokeYes, you just query ubuntucloudimageversiond. ;)11:29
Odd_Blokepitti: /etc/cloud/build.info11:30
pittibuild_name: server11:30
pittiserial: 20150328.111:30
pittiroot       464  0.0  0.3  23468  7784 ?        Ss   13:28   0:00 dhclient -1 -v -pf /run/dhclient.eth0.pid -lf /var/lib/dhcp/dhclient.eth0.leases eth011:30
pittithat's for /etc/network/interfaces.d/eth0.cfg11:30
* pitti -> lunch, bbl11:30
Odd_Blokepitti: Hmm, I'm not seeing it running in the GCE image built from that cloud image; I'll see if there's anything GCE-specific we're doing that might be messing things up.11:33
Odd_BlokeBut lunch does sound like a good idea.11:38
=== MacSlow|lunch is now known as MacSlow
=== Guest58714 is now known as gusnan
=== gusnan is now known as Guest42522
ari-tczewcan someone help with FTBFS on openblas? It's blocking other packages due to dep-waiting. https://launchpadlibrarian.net/200241214/buildlog_ubuntu-vivid-ppc64el.openblas_0.2.12-1ubuntu1_BUILDING.txt.gz12:26
ari-tczew /usr/bin/ld: ../libopenblas_power6p-r0.2.12.a(dcabs1.o): ABI version 1 is not compatible with ABI version 2 output12:26
=== Guest42522 is now known as gusnan
=== _salem is now known as salem_
tjaaltonso using a tmpfs overlay dir for sbuild seems to break with systemd or so, since it mounts a new tmpfs on top of /dev/shm every time sbuild is run :)12:49
Odd_Blokepitti: How are you launching your system with running dhclient?  I haven't managed to get a running dhclient in any of the environments I've tried.12:56
pittiOdd_Bloke: just with qemu, nothing extraordinary12:59
pitti$ kvm -net nic,model=virtio -net user -snapshot /srv/vm/adt-vivid-amd64-cloud.img13:01
pitti(I usually have some ssh port redirection, -m, virtio drive etc., but those should hardly matter, and it works like this13:01
pittikvm -snapshot /srv/vm/adt-vivid-amd64-cloud.img13:02
pittiworks with this, too13:02
cyphermoxgood morning!13:08
mvoRiddell: hi, sorry for the slow reply, I was trying to reproduce this in a chroot, but failed to trigger the bug with apt-get dist-upgrade --simulate, I try again with a real upgrade now (still chroot). if that does not trigger it I need to create a kubuntu install13:11
Riddellthanks mvo13:11
Riddellif I need to just remove the baloo4 package I can probably do that fine13:11
mvoRiddell: I have no idea right now, sorry. definitely a apt bug though13:12
mvoRiddell: still, we probably want to workarond it once its better understood13:12
happyaronmvo: "workarond" get my irssi highlights me, lol13:44
mvoRiddell: fun, I can not reproduce in a chroot with ubuntu-{minimal,standard}, kubuntu-desktop installed, looks like i need to try harder (i.e. upgrade works just fine)13:44
mvohappyaron: haha13:44
Riddellmeh13:47
* zyga wonders how to use the company vpn alongside local dns server13:51
ogra_very carefully ...13:51
argeshallyn: re: bug 1432644, i may need to try a completely clean vivid install, this is my laptop that's been incrementally upgraded since utopic. but at the moment i'm unable to use uvtool to install VMs and that error message was all I had to go by.13:58
ubottubug 1432644 in libvirt (Ubuntu) "VM permanently tries to read /dev/shm/lttng-ust-wait-5" [High,Fix released] https://launchpad.net/bugs/143264413:58
argesi'll upgrade to the latest libvirt and re-try13:59
tjaaltonhmm, commenting /dev/shm out from /etc/schroot/sbuild/fstab fixes the tmpfs overlays, wonder if that's a proper fix though14:00
jamespagearges, hallyn: let me know if you want me to disable the lttng integration in ceph - it appears to be making more problems that it solves right now14:07
shadeslayerogra_: ping14:08
ogra_shadeslayer, yo14:08
shadeslayerogra_: can we get a armhf tarball going on for Kubuntu vivid?14:08
shadeslayeror whom should I talk to about that14:08
shadeslayerI wrote this generic tool that can make RasPi2 images14:08
ogra_perferably someone from foundations, you will need to send patches fro the build system though14:09
shadeslayerhm ...14:09
shadeslayerogra_: https://github.com/rpi2-stuff/image , maybe you'd be interested in it for Ubuntu14:09
ogra_unless oyu only want a rootfs tarball14:09
shadeslayerhttps://github.com/rpi2-stuff/image/blob/master/config/vivid.yml14:09
shadeslayerogra_: I only want a rootfs tarball14:09
argesjamespage: I'll let hallyn decide, but I don't use the LTTng integration for ceph and it seems like something that should be optional since it would mostly be used for developers14:09
ogra_shadeslayer, that should be relatively straightforward14:10
shadeslayerright, so whom do I talk to :)14:10
ogra_mvo, ^^^ who does image stuff for distro nowadays ?14:10
argeshallyn: also my problem wasn't starting VMs it was creating a _new_ VM via 'uvt-kvm create' maybe try that too14:10
ogra_(i know you do snappy ... but i guess there is someone for general distro and flavour stuff)14:11
jdstrandtyhicks: hey, what is the status of cache loading for vivid?14:51
jdstrandtyhicks: I ask cause if it isn't going to land, someone needs to fix this:14:51
jdstrand1 processes are unconfined but have a profile defined.14:51
jdstrand   /sbin/dhclient (634)14:51
tyhicksjdstrand: we're still trying to nail down a detail in the libapparmor API14:52
tyhicksjdstrand: I'm working on that now14:52
tyhicksjdstrand: then the systemd changes need to happen14:52
tyhicksjdstrand: I'm thinking that it is all happening a bit too late to land in Vivid14:52
jdstrandtyhicks: I think I agree-- but we should land it as soon as 'w' opens14:53
tyhicks(when I said that "I'm working on that now", I meant that I'm working on the libapparmor API details and not the dhclient bug)14:53
tyhicksjdstrand: yeah, it'll definitely be ready then14:53
tyhicksjdstrand: do you have the dhclient bug # handy?14:55
jdstrandpitti: I'm filing it now14:55
jdstranderr14:55
jdstrandtyhicks: ^14:55
tyhicksoh14:55
jdstrandtyhicks: fyi, I updated bug #1385414 to reflect what you said above. filed bug #1438249 for dhclient15:03
ubottubug 1385414 in AppArmor "provide systemd compatible cache loading library" [Critical,In progress] https://launchpad.net/bugs/138541415:03
ubottubug 1438249 in systemd (Ubuntu) "/sbin/dhclient is unconfined after switch to systemd (aka, equivalent of upstart's network-interface-security.conf not implemented)" [Critical,Triaged] https://launchpad.net/bugs/143824915:03
jdstrandtyhicks: I'm not sure what the best solution is for bug #1438249-- perhaps discuss with pitti?15:04
jdstrandpitti: hi! :)15:04
tyhickspitti: hello - if you have any initial thoughts on how to address 1438249, please let me know. I'll come up to speed on what network-interface-security.conf does in the meantime.15:06
pittihey tyhicks15:09
pittityhicks: I'll have a look and respond to the bug15:10
awe_seb128, hey question for you... every so often on utopic, my desktop freezes during a hangout.  I have a working cursor, and audio, but everything is frozen such that I need to hard-power off my system15:31
awe_any suggestions for where I should file and bug and/or debug the problem?15:31
seb128awe_, hey, can you go to a vt in those cases?15:31
awe_hhhh, didn't try that...15:32
seb128I would start there15:32
seb128if you can, try to kill compiz15:32
seb128it might the wm being locked15:32
awe_everything else is frozen hard thought ( indicators, window controls, ... )15:32
awe_compiz still lives???15:32
seb128right15:32
awe_sigh...15:33
seb128could be xorg15:33
seb128could be compiz15:33
seb128could be the input stack/kernel15:33
* awe_ remember his first Canonical all-hands-meeting where we did a mock technical board session about including compiz by default15:33
awe_seb128, ack15:33
awe_thanks for the suggestion15:34
seb128yw!15:35
pittityhicks, jdstrand: responded to the bug, WDYT?15:45
shadeslayerogra_: btw out of curiosity, any particular reason why ubuntu-standard isn't installed in the core armhf tar?15:47
ogra_shadeslayer, to big ... not needed for embedded15:47
shadeslayerI see15:47
ogra_we dont use it in the phone images either15:48
shadeslayeroh I see15:48
* shadeslayer checks if his modifications work15:48
ogra_(which costs a price though ... extra maintenance and seed management for bit we actually want)15:48
ogra_*bits15:48
shadeslayertrue enough15:48
shadeslayerogra_: my script adds a user to the image it builds, but turns out there is no sudo on the rootfs15:49
shadeslayerso I can't effectively do much :p15:49
rbasakpitti: strikov and I are confused about bug 1409639. Why do you say Fix Committed in Vivid?16:05
ubottubug 1409639 in juju-core (Ubuntu Vivid) "juju needs to support systemd for >= vivid" [High,Fix committed] https://launchpad.net/bugs/140963916:06
pittirbasak: the fix landed in trunk apparently?16:21
jdstrandpitti: sorry was in a meeting. responded in the bug16:24
rbasakpitti: trunk upstream, but it won't be landing in Ubuntu for a while. Needs a major version bump for a version that isn't released yet.16:29
rbasakpitti: so I think upstream task Fix Committed makes sense, but not Ubuntu task as we'll be doing uploads that don't include the support in the interim.16:29
pittirbasak: ok; I don't mind much  (that's how I use "fix committed" -- as soon as the fix lands in trunk)16:58
rbasakpitti: I wasn't aware of that use. No worries. I don't mind either.17:00
rbasak(now I understand that use exists :)17:00
pittirbasak: please use the states as you usually do within juju, I don't want to step on your toes17:01
* pitti waves good night17:01
mvoRiddell: still no luck, did a fresh 14.10 kubuntu install in a VM, then upgraded to vivid (amd64), no error17:08
Riddellmvo: um really?17:10
Riddellmvo: I'll give it another go tomorrow, maybe it magically fixed itself since beta came out17:10
=== gusnan is now known as Guest68351
=== Guest68351 is now known as gusnan
=== gusnan is now known as Guest25333
mvoRiddell: hm, maybe I'm wrong, it looks like its incomplete, let me try again17:18
=== Guest25333 is now known as gusnan
=== gusnan is now known as Guest1649
=== Guest1649 is now known as gusnan
mvoRiddell: same result, 14.10 kubuntu/amd64 -> 15.04 dist-upgrade, no error, sorry18:42
infinitymvo: Did you mean to add ubuntu to a bunch of groups in that livecd-rootfs upload?18:59
infinitymvo: The changelog doesn't mention it.18:59
mvoinfinity: yes, sorry, I should have documented that, I can double check tomorrow though19:03
infinitymvo: Well, I'm less fussy about the changelog mentioning it (though that would be nice) and was mostly concerned that it was an accident, since adding a user to a couple of priviledged groups seems like the sort of thing one wouldn't want to do by accident. :P19:04
mvoinfinity: oh, I see. the ubuntu user is already part of sudo,docker this is just moving it into a different place, there is a existing 02-add_user_to_groups.chroot that does that. my mistake for not documenting that properly :/19:07
infinitymvo: Oh, so did this make 02-add_user_to_groups.chroot obsolete, or do you now do it twice? :P19:08
* infinity also thinks that syslog user's $HOME is rather suspiciously wrong, but that's not your fault.19:09
mvoinfinity: *cough* twice!19:09
* mvo adds a todo to fix that19:09
infinitymvo: *smirk*.  Alright, letting it in anyway, assuming that's a no-op and doesn't break.19:10
mvoinfinity: its tested in our PPA, so it should not break(tm) :)19:10
infinity(base)root@cthulhu:~# adduser adconrad sbuild19:10
infinityThe user `adconrad' is already a member of `sbuild'.19:10
infinity(base)root@cthulhu:~# echo $?19:10
infinity019:10
infinityYeah, looks like it'll work fine.19:10
infinityJust a pointless extra bit.19:11
infinitypitti: Can we, I dunno, make the systemd autopkgtests actually pass?  You're setting a bad example for everyone else. :P19:23
shadeslayermvo: who's in charge of the foundation bits for flavor images ?20:10
=== salem_ is now known as _salem

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!