[10:26] <jibel> can anyone look at bug 1020574, last working desktop image was June 29th
[10:26] <ubot2> Launchpad bug 1020574 in ubiquity "SystemError: Broken pipe while installing language packs" [Critical,Confirmed] https://launchpad.net/bugs/1020574
[10:31] <cjwatson> jibel: xnox was looking at that yesterday
[10:32] <xnox> jibel: still working on it.
[10:33] <xnox> cjwatson: langpack installation suppose to catch and ignore SystemError, but for some reason sys.excepthook is called, after the SystemError has bubbledup in a loop....
[10:34] <jibel> ok thanks. xnox I assigned the bug to you, so we know you're taking care of it.
[10:34] <cjwatson> distraction; the SystemError shouldn't be occurring here in the first place
[10:34] <xnox> cjwatson: then why the original code from ages ago has comment "ignore SystemError from langpack installation?"
[10:34] <xnox> ok
[10:36] <cjwatson> we didn't want random language pack failures to break the whole system; but that was more for things like package failed to unpack or something, not this sort of thing
[10:36] <xnox> ok
[13:16] <stgraber> xnox: can you easily reproduce that crash? if not, I have a pandaboard that's crashing every single time, so reproducing is easy here :)
[13:18] <xnox> stgraber: oh, it is reproducible every time with quantal daily since 29th in VM/desktop
[13:18] <stgraber> ok :) I only did arm installs recently, good to know any install will fail just as well :)
[13:19] <xnox> stgraber: so far my strace-foo didn't actually show me SIGPIPE, only the write("E: Broken Pipe") which comes from printing the stack trace in python, way to late.
[13:19] <xnox> stgraber: I can't wait to install my panda =)
[13:19] <xnox> stgraber: it's reproducible in a vm, which is what I am using now.
[13:23] <cjwatson> it might be EPIPE not SIGPIPE
[13:23] <cjwatson> in fact I'd say that's likely given that we're seeing several of them
[13:24] <xnox> cjwatson: not present in my strace either. I attached to the process just before clicking install now, cause otherwise the strace log was 300MB due to printing a line for almost each character of a debconf template fetch.
[13:25] <xnox> cjwatson: I am doing a full one by executing ubiquity under strace and logging into a separate hard-drive now.
[13:25] <xnox> also no obvious changes in python-apt nor ubiquity around the time.
[13:38] <cjwatson> apt and python-apt both had major changes at the relevant time
[14:02] <brendand> we have a bug with updating precise, which only happens when running the upgrade from a preseed: https://bugs.launchpad.net/checkbox-satellite/+bug/1021293
[14:02] <ubot2> Ubuntu bug 1021293 in checkbox-satellite "Ubuntu 12.04 install stalls when doing apt-get upgrade" [Undecided,New]
[14:06] <cjwatson> followed up
[14:20] <brendand> cjwatson, funny you ask. we've observed 'apt-get <defunct>' in ps
[14:20] <brendand> cjwatson, i'll ask the reporter to give full details though
[14:20] <cjwatson> yeah, I'm not going to get anything much out of one line
[14:20] <brendand> asap
[14:23] <roadmr> brendand, cjwatson: bug updated with the requested process list
[14:26] <cjwatson> hmph.
[14:26] <cjwatson> not obvious, will have to strace
[14:55] <cjwatson> brendand,roadmr: hm, worked for me in a normal session with strace attached
[14:55] <brendand> cjwatson, you mean without the preseed?
[14:55] <cjwatson> oh, admittedly I'm using a newer image
[14:56] <cjwatson> with the presee
[14:56] <cjwatson> d
[14:56] <brendand> cjwatson, a newer image - of precise?
[14:56] <cjwatson> yeah
[14:56] <xnox> brendand: there are daily precise images published with proposed enabled.
[14:57] <cjwatson> might be coincidental though and in any case wouldn't be a reason not to fix the underlying bug
[14:57] <cjwatson> probably just means I avoid some upgrade that triggers the bug, since it's already in the squashfs
[14:57] <brendand> xnox, details?
[14:58] <cjwatson> http://cdimage.ubuntu.com/precise/daily/ - but they aren't guaranteed, you shouldn't be relying on them for anything any more than you would rely on random daily builds
[14:58] <cjwatson> sorry, .../daily-live/
[14:58] <brendand> cjwatson, right. interesting to know they are there though
[14:59] <brendand> cjwatson, it would be good to know if it reproduces for you using the release image.
[14:59] <cjwatson> I'm rsyncing back
[14:59] <brendand> cjwatson, by the way, this doesn't happen on servers
[15:00] <cjwatson> yeah, I know, not surprising at all
[15:00] <cjwatson> but not very helpful either TBH :)
[15:00] <cjwatson> different installer, so ...
[15:01] <cjwatson> (and ubiquity/success_command is already quite specific to ubiquity)
[15:02] <brendand> cjwatson, right - but the bug report is just to trim the symptoms down. the servers also use a preseed with the same steps - so the update is run in late_command
[15:03] <cjwatson> I know, and fair enough, but it's not a big help in this case
[15:03] <cjwatson> although granted, if a similar symptom had appeared on server, then it would probably have been quicker to diagnose there
[15:04] <brendand> cjwatson, if you need us to do anything, ping me or roadmr
[15:04] <cjwatson> ETA 40 mins or so for the image to arrive, so might not be today
[15:52] <mpt> I think ... I am going to suggest removing the big pictogram icons from the "Installation type" step
[15:52] <mpt> Partly to save space, and partly because they just aren't that expressive
[15:52] <mpt> (Saving space to make room for "[ ] Encrypt the new Ubuntu installation for security")
[19:41] <Blackyugin> Hi all, I work to ubiquity and I want to create my DBus witch connect to Ubuntu One. I see oneconf's DBusconnect and I need to know service's name. In oneconf is "com.ubuntu.OneConf". How I can know the Ubuntu One's name and all informations (hosts_object_name, package_set_interface, host_interface, etc.). Thanks a lot