[15:00] <pitti> o/
[15:01] <infinity> \o
[15:01] <barry> o/
[15:02] <robru> yo
[15:02]  * slangasek waves
[15:03] <slangasek> #startmeeting
[15:03] <meetingology> Meeting started Thu Jun 25 15:03:40 2015 UTC.  The chair is slangasek. Information about MeetBot at http://wiki.ubuntu.com/meetingology.
[15:03] <meetingology> Available commands: action commands idea info link nick
[15:03] <slangasek> [TOPIC] Lightning round
[15:04]  * pitti can hear slangasek's dice rolling
[15:04] <slangasek> $ echo $(shuf -e barry doko stgraber bdmurray slangasek caribou infinity sil2100 robru cyphermox pitti)
[15:04] <slangasek> caribou slangasek stgraber barry cyphermox sil2100 infinity pitti doko robru bdmurray
[15:04] <slangasek> that's the right list this time, yes? :)
[15:04] <infinity> Close enough.
[15:05] <pitti> missing mvo and jhunt :)
[15:05] <infinity> :(
[15:05] <pitti> or, rather, :(
[15:05] <infinity> Yeah, and that xnox guy too.
[15:05] <infinity> Your list sucks.
[15:05] <slangasek> heh
[15:05] <xnox> yo
[15:05] <infinity> (And cjwatson too...)
[15:05] <slangasek> caribou: hi, anything for us this morning?
[15:05] <xnox> pitti: missing mvo?
[15:06] <xnox> pitti: at least you are in the right team now ;-)
[15:06] <infinity> xnox: He got reorged.
[15:06] <ogra_> xnox, snappy ate him
[15:06] <pitti> xnox: he moved to the snappy team
[15:06] <xnox> infinity: ogra_: at least pitti is here now ;-)
[15:06] <xnox> anyway, working with doko to sort out gcc5 in debian
[15:06] <xnox> ..
[15:07] <slangasek> haha
[15:07] <slangasek> xnox: thanks ;)
[15:08] <slangasek>  * hiring: progress on the Java role,
[15:08] <slangasek>  * looking into bug #1464442, which seems to be a strange corner case in upstart on livefs
[15:08] <slangasek>  * working with cyphermox on grub uefi --removable search path, for snappy personal
[15:08] <slangasek>  * getting conference travel all sorted out for the summer
[15:08] <slangasek> stgraber is out this morning
[15:09] <slangasek>  * sick this week, running at about 3/4 capacity
[15:09] <slangasek> (done)
[15:09] <slangasek> barry:
[15:09] <barry> debuntu: zope.interface 4.1.2-1, zope.i18n 4.0.1-1, zope.testing 4.2.2-2
[15:09] <barry> Python 3.5 transition test rebuild work
[15:09] <barry> python-pip and related SRUs for trusty; should be all sorted out now
[15:09] <barry> trying to reboot the DPMT git-dpm conversion work
[15:09] <barry> --done--
[15:10] <slangasek> cyphermox:
[15:10] <cyphermox> hello!
[15:10] <cyphermox>  * (before vacation) ppc64-diag update (bug 1459698)
[15:10] <cyphermox>  * catching up with what I missed
[15:10] <cyphermox>  * testing multipath-tools 0.5.0
[15:10] <cyphermox>    - it's not collaborating with qemu so far...
[15:10] <cyphermox>  * bug 1460521: UEFI BootOrder not empty after efibootmgr -B
[15:10] <cyphermox>  * debugging snappy personal UEFI boot (bug 1468111)
[15:10] <cyphermox>    - related to bug 1097570 which I was already working on
[15:10] <cyphermox>  * Wednesday: national holiday
[15:10] <cyphermox>  * discussing NM/wpa update plans for wily with Tony.
[15:10] <cyphermox>  * more grub fixes:
[15:11] <cyphermox>    - don't block with 'diskfilter writes are not supported' (bug 1274320)
[15:11] <cyphermox>    - OsIndicatorsSupported '1f' fix still to land.
[15:11] <cyphermox>    - arm64-signed images from dannf.
[15:11] <cyphermox> (done)
[15:12] <infinity> Oh, sil2100 isn't here.
[15:12] <infinity> - Managed a few transitions in wily
[15:12] <infinity> - Wrangled an emergency kernel SRU
[15:12] <infinity> - Sorted out d-i for the 4.0 kernel (pending in the kteam PPA)
[15:12] <infinity> - General AA/SRU work
[15:12] <infinity> - Rebuilt the ppc and ppc64el buildd VMs
[15:12] <slangasek> no sil2100 it seems, so infinity
[15:12] <infinity> - Got the 14.04.3 betas building
[15:12] <infinity> - Helping with Wily Alpha 1
[15:12] <infinity> (done)
[15:12] <infinity> (note that Canadians have July 1 off next week)
[15:12]  * xnox trolls
[15:12] <xnox>  * onboarding jodh in the team
[15:12] <xnox> ..
[15:13] <pitti> systemd: package/test version 221 (held back because of alpha-1), various bug fixes, postinst migration for old mountall fstab options
[15:13] <pitti> autopkgtest in cloud:
[15:13] <pitti>  - implement result fetching from swift, teach debci web UI to get along with artifacts in swift (all upstream now)
[15:13] <pitti>  - some autopkgtest improvements for running in cloud
[15:13] <pitti>  - create charms for debci web UI and workers, and deployment script; the full thing can now be brought up with a single command
[15:13] <slangasek> xnox: :P
[15:13] <pitti>  - sort out various permission/public IP/DNS name/image import issues with IS; http://autopkgtest.ubuntu.com now (mostly) works and displays actual results
[15:13] <pitti> misc: release new vivid -base langpack refresh, for updated ubuntu-docs
[15:13] <pitti> PLAN:
[15:13] <pitti>  - discuss/fix rabbit access from britney with IS
[15:13] <pitti>  - mojo-ify autopkgtest cloud setup (requirement for DevOps)
[15:13] <pitti>  - land systemd 221
[15:13] <pitti> ␃
[15:13] <cyphermox> pitti: need to be more inventive than ␃ :)
[15:14] <slangasek> doko:
[15:14] <robru> no doko tho?
[15:14] <pitti> cyphermox: sorry, second week only; I still have some leeway for creative EOD markers :)
[15:14] <pitti> 'T even
[15:15] <cyphermox> ;)
[15:15] <slangasek> should be a doko, he's gone yesterday+tomorrow and here today
[15:15] <pitti> . o o { must be summer time! }
[15:15] <doko> - two GCC backports upstream
[15:15] <doko> - updated GCC packages
[15:15] <doko> - raised severity of GCC issues which are fixable without GCC 5 being the default
[15:15] <doko> - prepared d-d-a (and u-d GCC mail)
[15:15] <doko> (done)
[15:15] <slangasek> tada
[15:15] <slangasek> robru:
[15:15] <doko> no, no, I'm here ...
[15:15] <robru> * CI Train:
[15:15] <robru>  - Rewrote prepare_silo.py (+553/-714), no feature regressions, bringing test coverage from 10% to 100%
[15:15] <robru>  - Overall test coverage now up to 98%.
[15:15] <robru>  - Implemented PPA discovery so that the silo count is no longer hard-coded.
[15:15] <robru>  - Add '-sa' to source package build to fix barry's system-image uploads.
[15:15] <robru> * CI Train Charm:
[15:15] <robru>  - Add ppa.launchpad.net to known_hosts so that dput over sftp works out of the box.
[15:15] <robru> * Bileto:
[15:15] <robru>  - Integrated dashboard so that staging & production can have independent ones.
[15:15] <robru> * Bileto Charm:
[15:15] <robru>  - rsync silo statuses in from jenkins to support dashboard integration
[15:15] <robru> * Mojo Spec:
[15:15] <robru>  - Force storage relations to be set before other relations; fixes race condition
[15:15] <robru>  - A few iterations on the relation between bileto and jenkins.
[15:16] <slangasek> +1 for -sa
[15:16] <bdmurray> robru: done?
[15:16] <slangasek> bdmurray:
[15:17] <robru> yeah done
[15:17] <bdmurray> updated retracer-dependencies to depend on dpkg-dev
[15:17] <bdmurray> submitted RT to have retracers access launchpad (RT 82159)
[15:17] <bdmurray> landed overlay PPA retracing support on production retracers
[15:17] <bdmurray> investigation into overlay PPA retracing success and failure
[15:17] <bdmurray> research into maliit-framework and qmlscene retracing failures
[15:17] <bdmurray> submitted apport merge proposal fixing bug LP: #1468154
[15:17] <bdmurray> review / testing of darkxst's apport ppa branch
[15:17] <bdmurray> replied to RT regarding removing core files from swift
[15:17] <bdmurray> improved phased-updater to better handle errors when communicating with errors
[15:17] <bdmurray> fixed a traceback with the phased-updater
[15:17] <bdmurray> ✔ done
[15:17] <bdmurray> foundations bug triage
[15:18] <slangasek> great
[15:18] <slangasek> any questions over status?
[15:19] <infinity> slangasek: Your upstart bug is disconcerting.
[15:19] <slangasek> infinity: innit though?
[15:19] <robru> slangasek: bdmurray: http://imgur.com/Kv1wsa3 is my paste broken? why do people often ask if I'm done?
[15:19] <infinity> slangasek: cking and smb were noting some fun /bin/sh /proc/fd fork magic in their traces that I've only ever seen when I have my "runlevels run twice on boot" bug, I wonder if that's coincidence, or if it's the same bug somehow.
[15:20] <slangasek> robru: heh, your dinner was intercepted en route
[15:20] <robru> slangasek: was that the only line missing?
[15:20] <cyphermox> yes
[15:20] <robru> hmm
[15:20] <cyphermox> robru: was it a notice?
[15:20] <pitti> robru: we didn't get the "done like dinner" line
[15:21] <robru> cyphermox: nope, just all one big paste. IRC client put each line as a separate message as far as I can see from my side.
[15:21] <cyphermox> might be some flooding prevention thing
[15:22] <robru> grumble
[15:22] <slangasek> infinity: yep, I'll keep picking away at that upstart bug... good news though, it's not an eglibc bug ;)
[15:22] <ogra_> it was the "no flooding with dinner in ubuntu channels" rule
[15:22] <slangasek> [TOPIC] AOB
[15:22] <slangasek> anything else?
[15:23] <pitti> does anyone have experience with putting rabbitmq into a cloud and exposing it?
[15:23] <pitti> (wrt. authentication and stuff)
[15:23] <robru> pitti: ci team knows all about that
[15:23] <pitti> robru: ack; I'll ask cprov
[15:23] <robru> good call
[15:24] <slangasek> and hey, we have a new sbuild, which means we have a new edk2 package, which means we can have arm64 cloud instances
[15:24] <pitti> (the standard charm store charm isn't very configurable, and even seems to open up a default guest:guest user remotely..)
[15:24] <slangasek> as soon as someone wires that up to openstack and shows that it works ;)
[15:24] <pitti> \o/
[15:24] <infinity> slangasek: It couldn't possibly have been a glibc bug, but I don't know why I don't immediately blame init everytime someone blame my postinst.  It's pretty much always init's fault. :P
[15:24] <slangasek> infinity: heh
[15:25] <pitti> infinity: init's or one of its child processes, yes :)
[15:25] <infinity> pitti: Well, in the case of glibc's postinst, init specifically, since it reexecs.
[15:26] <infinity> pitti: Except that I don't currently reexec systemd, cause someone talked me into not doing so back when it was new and non-default.  Should revisit that.
[15:26] <pitti> "systemctl daemon-reexec" works fine
[15:26] <infinity> pitti: Indeed.  And tends to happen often enough anyway, since I added that trigger. :P
[15:26] <pitti> or SIGTERMing pid 1
[15:27] <infinity> pitti: But still, would be proper to reexec when deps change to get security updates, etc.
[15:27] <infinity> (As I do for sysv and upstart)
[15:27] <pitti> *nod*
[15:30] <slangasek> #endmeeting
[15:30] <meetingology> Meeting ended Thu Jun 25 15:30:16 2015 UTC.
[15:30] <meetingology> Minutes:        http://ubottu.com/meetingology/logs/ubuntu-meeting/2015/ubuntu-meeting.2015-06-25-15.03.moin.txt
[15:30] <slangasek> thanks, everyone!
[15:30] <barry> thanks!
[15:30] <infinity> slangasek: Oh, I've been looking into some patches we should backport from linus's trunk to get full armv7 insn emulation on armv8 (and another patch I need to write), so we're probably a couple of SRU cycles away from armv7-on-armv8 kvm buildds.
[15:30] <pitti> thanks!
[15:30] <infinity> slangasek: But it'll probably be that long until armv8 in scalingstack is happy anyway.
[15:30] <cyphermox> thanks!
[15:30] <slangasek> infinity: I thought our arm64 hardware is fully armv7 capable anyway
[15:30] <cyphermox> robru: missing lines> it's why I copy things just three to five lines at a time, old habit from copy-pasting a humongous Cisco FWSM rule-set that would otherwise crash the device... with fun consequences.
[15:31] <slangasek> (it's not a requirement of the ABI, but it's an option that some vendors implemented)
[15:31] <infinity> slangasek: Yeah, so, that's the ARM company line.  It's a lie.  v8 is missing a few v7 instructions that are emulated in software on the latest 4.x kernels. :P
[15:31] <robru> cyphermox: strange though that the only missing line was the one starting with a (. I'll try not using that next time...
[15:31] <cyphermox> hahah
[15:31] <infinity> slangasek: And those instructions are missing from every implementation in the wild, AFAIK, including ours.
[15:32] <slangasek> infinity: are you talking about the pre-thumb2 instructions that we're not using anywhere in the distro?
[15:32] <infinity> slangasek: Basically, there were some v6 instructions that were marked "deprecated" in v7, and ARM decided that those didn't "count" when claiming v8 was v7 compat.
[15:32] <slangasek> infinity: also, shouldn't qemu do this instead of the kernel?
[15:32] <slangasek> right
[15:32] <slangasek> so they were already deprecated and AIUI we didn't use them... doesn't seem to me that it should have a practical impact
[15:33] <infinity> slangasek: The SWP instructions.  And sure, qemu would do it if we were doing armv7 qemu on armv8 hosts, but that also doesn't work yet.  We'd be doing v7 chroots in v8 guests.
[15:33] <slangasek> ok
[15:34] <infinity> slangasek: Anyhow, yes, probably no practical impact, but I'd rather be fully v7 compat, since we can.
[15:34] <slangasek> infinity: ok.  but I don't think this should be a blocker for scalingstack
[15:35] <infinity> slangasek: No, not a blocker for scalingstack PPAs.  Maybe a blocker for ditching kishi, since I don't want any regression potential that we know about in advance.
[15:35] <infinity> slangasek: Either way, I'll get it sorted soon, so meh.