=== _salem is now known as salem_ === salem_ is now known as _salem === chihchun is now known as chihchun_afk === chihchun_afk is now known as chihchun === chihchun is now known as chihchun_afk [04:21] Good morning === _salem is now known as salem_ === salem_ is now known as _salem === chihchun_afk is now known as chihchun [07:26] pitti, did you see http://d-jenkins.ubuntu-ci:8080/job/utopic-adt-killbots/1/ARCH=amd64,label=adt/console ? [07:26] jibel: haven't looked at that one yet; I fixed the autopkgtest bug with eglibc this morning, though [07:27] jibel: oh, there's a reasonable chance that this works now, I updated to latest git this morning [07:27] jibel: and I now have test cases/fixes for UTF-8 control stuff under LANG=C [07:27] * pitti jumps through login stuff again to retry -- argh slow 2G network in train :) [07:28] pitti, ok, thanks. I'll restart the job [07:28] jibel: thanks [07:30] jibel: ça semble mieux \o/ [07:31] pitti, yes, 1 more green \o/ [07:32] *phew*, how could we live without killbots? [07:34] jibel: btw, how's your ssh runner coming along? do you need a hand with this? [07:35] jibel: after the recent Tests class refactorization I'm going to add support for parsing .click packages [07:36] pitti, the adt side is okay I think, it needs tests and documentation. I'm now writing the setup script for lxc as an example. but I must finish the work on hwe eol notifications this week before I can continue on adt [07:36] for now I'll test them in a hand-crafted schroot, but I'll soon need to test them in ubuntu-emulator or real iron === qwebirc888141 is now known as slickymasterWork === chrisccoulson_ is now known as chrisccoulson === vrruiz_ is now known as rvr [10:22] pitti, did you already provisioned a precise VM with adt-buildvm-ubuntu-cloud ? [10:22] I tried and it fails with a timeout [10:25] jibel: no, I'm afraid I never tried that; just trusty and utopic [10:25] jibel: can you please file a bug? I'm happy to look into that once I'm back on a place with real bandwidth [10:25] jibel: (LP is fine, no need to wrangle with the Debian bug tracker :)) ) [10:27] pitti, I will === chihchun is now known as chihchun_afk [11:33] jibel: btw, did you see my q about helping with virt-ssh? [11:43] pitti, yes, I said that the driver is working, and I'm now on a setup script for lxc as an example, but I've to finish hwe eol notification tests before that. [11:43] jibel: oh, cool! thanks [11:43] jibel: right, was just wondering if you stumbled over other weird adt-run bugs which I should fix in the meantime [11:44] pitti, no weird bug so far :) [11:46] pitti, I reported bug 1330950 , for the provisioning error of Precise VMs [11:46] bug 1330950 in autopkgtest (Ubuntu) "adt-buildvm-ubuntu-cloud fails to provision Precise VMs with timeout in cloud-init" [Undecided,New] https://launchpad.net/bugs/1330950 [11:47] jibel: thanks; probably just some missing feature/bug in older cloud-init which it needs to deal with [11:49] jibel: that blocks you for the dkms tests? [11:55] pitti, that blocks me on hwe eol tests. I didn't want to use the old provisioning script and must test in VM to boot with the new kernel and fake system time [11:56] I think we need to special case Precise [11:56] jibel: ack; I'll have bandwidth again this afternoon/evening (but will be out to visit my grandma in the afternoon) [11:57] pitti, don't worry I'll have a look [11:57] jibel: can you use a manually built VM (from prepare-testbed, with purging cloud-init) until that? [11:57] pitti, of course, it is not urgent [11:58] jibel: well, I'll fix it by tomorrow for sure === _salem is now known as salem_ [13:01] jibel: oh, thanks for the patch! [13:05] pitti, now cloud-init accepts cloud-config but run_cc_modules fails with No module named cc_keys_to_console and no module named cc_phone_home [13:13] hm, that's because cloud-init doesn't enjoy that its modules are being purged under its feet [14:12] jibel: oh, is it being purged while it's not done yet? [14:12] jibel: downloading precise cloud image now, I'll have a look [14:13] pitti, yes, so the 3 modules keys_to_console, phone_home and finale_message are not executed [14:13] jibel: ok, so it's not waiting long enough, I suppose [14:14] pitti, but right now I'm writing a user-data file specific to Precise because write_files is not supported [14:14] replacing write_files by runcmd [14:14] jibel: ah, but runcmd works? I'll replace that then [14:14] pitti, it does [14:16] jibel: ah, I see -- so I shouldn't purge cloud-init in runcmd, as during that time it's still running [14:16] jibel: but instead should wait until it's done, and then purge/shutdown [14:19] pitti, right. and probably do a shutdown +1 instead of 'now' [14:19] final_message must be executed after the last runcmd [14:20] pitti, do you want me to replace write_files by runcmd or you do? [14:20] jibel: I'm currently converting it [14:20] pitti, k [14:21] jibel: I haven't yet checked how to see when cloud-init is done; do you already know an official way? === Ursinha is now known as Ursinha-afk [14:22] pitti, when the hook cc_final_message.py is executed [14:23] pitti, the file /var/lib/cloud/instance/boot-finished is created with a default message: [14:23] does that write a file ... ah! [14:23] "cloud-init boot finished at $TIMESTAMP. Up $UPTIME seconds" [14:23] jibel: so a runcmd could fork off a shell which polls for that file, purges cloud-init, and shuts down [14:23] pitti, the message is configurable in cloud-config [14:24] pitti, that would be perfect [14:24] jibel: I'd like to do everything from runcmd, that avoids having to actively ssh in and do stuff [14:25] agreed [14:25] jibel: do you know, can I put comments into the runcmd: section? [14:25] # do foo [14:25] - ... [14:25] etc? [14:27] pitti, not sure on Precise, but it seems to support multilines scripts even if it is not documented [14:29] hmm, < a ugly: echo -e "#!/bin/sh\n### BEGIN INIT INFO\n# Provides: autopkgtest\n# Required-Start:...">/etc/init.d/autopkgtest would work I guess [14:51] jibel: with a super-long line, yes; but I'm fighting with "Failure to shellify" cloud-init errors even for very simple files [14:51] * pitti pokes [14:51] plus, apt_upgrade: false doesn't seem to work, annoying [15:11] jibel: I tried putting the script onto the seed iso, but that's not mounted any more :/ [15:11] this is ridiculously complicated :/ [15:11] apparently I need to json-escape ':' [15:14] or whatever that file format is; as soon as I have a : in the echo, it breaks [15:18] pitti, it's yaml, and : is used for associative arrays [15:20] jibel: I got it, I can quote the entire command [15:29] jibel: so http://cloudinit.readthedocs.org/en/latest/topics/examples.html#call-a-url-when-finished is apparently also not yet supported in precise :/ [15:30] (power_state) [15:30] http://cloudinit.readthedocs.org/en/latest/topics/examples.html#reboot-poweroff-when-finished I mean [15:32] :( [15:34] a hack would be to create a script in a runcmd that do a "shutdown -P +1 & " and execute it [15:34] and purges cloud-init [15:35] jibel: yes, I'm testing this now: [15:35] - (while [ ! -e /var/lib/cloud/instance/boot-finished ]; do echo POLL; sleep 1; done; apt-get -y purge cloud-init; shutdown -P now) & [15:37] which kind of seems to work, I just don't get a serial terminal when booting the resulting VM [15:39] right, no serial terminal upstart job any more [15:39] in trusty I keep that [15:41] pitti, can you paste your diff? [15:42] jibel: I just pushed the write_files -> runcmd update [15:42] jibel: http://paste.ubuntu.com/7659030/ [15:42] thanks [15:42] I pulled it [15:42] jibel: that disables some stuff which makes the iteration go faster [15:42] AUTOPKGTEST_BASE=`pwd` tools/adt-buildvm-ubuntu-cloud --cloud-image-url=file:///home/martin/download/ubuntu -v -r precise -o /tmp --no-apt-upgrade [15:42] jibel: that uses the locally downloaded cloud image, and disables the dist-upgrade [15:43] jibel: so I mostly need to retain the serial console now, other bits seem to work fie [15:43] fine [15:43] I use the same command line [15:43] cp /etc/init/tty2.conf /etc/init/ttyS0.conf && [15:43] …………sed -i s,tty2,ttyS0,g /etc/init/ttyS0.conf 2>/dev/null && [15:43] …………log "enabled console on ttyS0" [15:43] jibel: that's in cloud-init-0.6.3/tools/uncloud-init [15:44] perhaps I just need to do that explicitly [15:45] but in utopic it's not just a simple s/tty2/ttyS0/ [15:48] jibel: I don't even know where utopic's ttyS0.conf comes from -- not from cloud-init [15:57] jibel: got it [15:58] adt-run [17:58:24]: ERROR: unexpected error: "dpkg --unpack /tmp/adt-run.RCYtYo/adt-satdep.deb" failed with stderr "dpkg-deb: error: `/tmp/adt-run.RCYtYo/adt-satdep.deb' is not a debian format archive [15:58] jibel: that might now be something xz related perhaps [16:00] jibel: my current diff: http://paste.ubuntu.com/7659114/ [16:01] jibel: hm no, I also get weird tar errors for simple packages; I'm afraid that's something I need to look at tomorrow [16:04] pitti, thanks, I'll continue a little bit. we'll probably want to move cloud-config outside of buildvm and load precise specific version for precise only and keep it as it was for T and U. The previous version was much nicer and legible. [16:05] jibel: I pushed both fixes now [16:05] jibel: now I don't have a local delta any more [16:06] jibel: I have no idea yet why virt-qemu's runcmd fails so badly on precise [16:08] jibel: oh, problems with 9p perhaps [16:13] pitti, likely, adt-satdep is 0 byte on the guest and 688 on the host [16:14] jibel: so calling something simple like "runcmd uptime" works fine [16:14] but stdin pipelining is b0rked [16:14] $ echo hello | /tmp/adt-virt-qemu.ashsl8a8/runcmd cat [16:14] $ [16:14] no output [16:15] jibel: interestingly, maybe this is just an aggravation of the race condition that we occasionally see in utopic as well [16:16] just that this one is perfectly reproducible for me [16:18] jibel: argh, *headdesk* [16:18] $ python3 --version [16:18] The program 'python3' is currently not installed. You can install it by typing: [16:19] heh [16:19] jibel: so, need to use python2 for eofcat, at least for 12.04 (or more correctly, detect what's available) [16:21] jibel: temporary hack for you to unblock you: http://paste.ubuntu.com/7659188/ [16:21] jibel: I'll fix this properly tomorrow [16:21] jibel: maintenant, c'est l'heure de courier :) [16:22] pitti, merci beaucoup et bonne course! === salem_ is now known as _salem === _salem is now known as salem_ === PaulW2U is now known as G4MBY === roadmr is now known as roadmr_afk === roadmr_afk is now known as roadmr [20:34] What's the bug number for the issue where the isolinux menu doesn't show for images not using maybe-ubiquity? [20:53] hi does anybody know a good guide to quality / testing? [20:55] gerry_: depends what you want. "clean code" is a good start [20:56] also "working effectively with legacy code" and finally "xUnit test patterns" for unit test focussed content [20:58] Hi Sorry that is far above my head at the moment I have just been reading the quality page on the ubuntu site and thought I would like to contribute [21:00] I what was written not very clear. I was searching for a step by step guide [21:01] gerry_: to what exactly - testing for us? or something else [21:01] try and be a bit more specific about how you think you can help :) [21:03] gerry_: did you see https://wiki.ubuntu.com/QATeam/Roles [21:03] sorry I am just raising my question from what I read on the quality team page I was reading about bug triaging but found the explanation not verz [21:04] sorry not very clear [21:04] oh right - I'll not be of much help to you about bug triaging :) [21:05] gerry_: did you see the triaging wiki page? https://wiki.ubuntu.com/Bugs/Triage [21:06] thats ok the link you supplied looks like what I was trying ineptly to ask [21:06] good :) [21:06] gerry_, elfy :-) [21:06] balloons: hi there :) [21:06] just imagine how much I'd get everyone into triaging :D [21:07] how much trouble [21:07] thank you :) [21:07] you're welcome :)