16:03 #startmeeting 16:03 Meeting started Thu Jan 28 16:03:05 2016 UTC. The chair is slangasek. Information about MeetBot at http://wiki.ubuntu.com/meetingology. 16:03 16:03 Available commands: action commands idea info link nick 16:03 [TOPIC] Lightning round 16:03 $ echo $(shuf -e barry doko bdmurray slangasek caribou infinity sil2100 robru cyphermox pitti tdaitx xnox chiluk) 16:03 caribou tdaitx cyphermox sil2100 pitti robru bdmurray slangasek chiluk doko xnox barry infinity 16:04 caribou: hi :) 16:04 tdaitx: ok, you're up :) 16:05 # 2016-01-28 16:05 * Updated OpenJDK 6 from 6u37-1.13.9 to 6u38-1.13.10 (based on IcedTea 1.13.10) 16:05 * Provided OpenJDK 6u38 backports to the security team for Wily, Vivid, Trusty, and Precise 16:05 * Porting Jean-Baptiste TCK scripts from JCK 6 to JCK 7 16:05 * Travelling today to Brussels for FOSDEM 16:05 (done) 16:05 - libaudit support: shadow and openssh (bug LP: #1478087) 16:05 - multipath-tools SRUs for 14.04.4: 16:05 - bug LP: #1432062 - spaces in device names 16:05 Launchpad bug 1478087 in lightdm (Ubuntu Vivid) "Add libaudit support" [Medium,Triaged] https://launchpad.net/bugs/1478087 16:05 Launchpad bug 1432062 in multipath-tools (Ubuntu) "multipath-tools-boot: support booting without user_friendly_names on devices with spaces in identifiers" [Medium,Confirmed] https://launchpad.net/bugs/1432062 16:05 - bug LP: #1526984 - readonly bindings for multipathd -B 16:05 - bug LP: #1503286 - let udev settle before mounting. 16:05 - bug LP: #1496210 - defualt values for IBM 2810XIV storage 16:05 Launchpad bug 1526984 in multipath-tools (Ubuntu) "ISST-LTE: root mpath device unavailable after installation" [Critical,In progress] https://launchpad.net/bugs/1526984 16:05 tdaitx, look up the beer location for tomorrow night 16:05 Launchpad bug 1503286 in multipath-tools (Ubuntu Trusty) "ISST-LTE: Boot of Ubuntu15.10 lpar fails: "mounting /dev/sdn2 on /root failed: Device or resource busy" [multipath]" [Critical,In progress] https://launchpad.net/bugs/1503286 16:05 Launchpad bug 1496210 in multipath-tools (Ubuntu Trusty) "multipath-tools lacks the default settings for IBM 2810XIV storage system" [Medium,In progress] https://launchpad.net/bugs/1496210 16:05 - debugging parted device naming policy and ioctl errors (bug LP: #1536008) 16:05 - reviewing MIRs: fwupd (bug LP: #1536871) - appstream (bug LP: #1538293) 16:05 (done) 16:05 Launchpad bug 1536008 in parted (Ubuntu) "ISST-LTE: parted command shows "device-mapper: table ioctl on failed: No such device or address" error" [Undecided,In progress] https://launchpad.net/bugs/1536008 16:06 Launchpad bug 1536871 in fwupd (Ubuntu) "[MIR] fwupd" [Undecided,New] https://launchpad.net/bugs/1536871 16:06 Launchpad bug 1538293 in appstream (Ubuntu) "[MIR] appstream" [Undecided,New] https://launchpad.net/bugs/1538293 16:06 oh, no sil2100 16:07 autopkgtest: 16:07 indeed 16:07 - Enable SSL on debci, fix some remaining issues in autopkgtest-retrier, and roll out self-service retry buttons 16:07 - Refine web-based retry to check uploader permissions 16:07 - Fix wrong kernel versions with apt pinning 16:07 - britney: Add option for working with a lot of shared silo instances relative to Ubuntu (#1537868) 16:07 - britney: Fix crash with NBS binaries in target release (reported by robru) 16:07 - (ongoing) more experiments with running armhf tests through remote LXD in Scalingstack; still blocked by "locks up after some time" bug 16:07 - (ongoing) Investigate missing proxy vars on ppc64el/trusty (#1539126) 16:07 distro: 16:07 - init-system-helpers: add autopkgtest 16:07 - init-system-helpers: review our remaining delta, reduce most of it (part of that: #1539016); we can sync after xenial 16:07 - merges: gnupg, ifupdown 16:07 - network-manager: Fix/test/apply Bryan's patch for fixing NFS mounts (#1515446) 16:07 - numexpr: Fix autopkgtest regression 16:07 - udev: Fix persistent net names with d-i for trusty (#1537136) 16:07 pitti: \o/ 16:07 - Clean up usage of /var/log/udev (#1537211) 16:07 - Review initramfs-tools, init-system-helpers, and sysvinit merges from Andy; systemd tests spotted a regression in update-rc.d, fix that 16:07 - Discuss apport implementation for crashes in container with stgraber; we have a solution now, tests needs to be written and then we can enable this again 16:07 END 16:07 lp:cupstream2distro 16:07 * fix traceback when abandoning silos 16:07 * fix erroneously setting published_versions excessively 16:07 * fix noisy transient errors in status setting 16:07 * set branches as merged after pushing to trunk 16:07 lp:bileto 16:07 * misc code cleanup 16:07 * disable some unused britney features 16:07 * fix when lander_signoff is automatically cleared 16:07 * add documenation/network topology chart 16:07 misc 16:07 * audit bugs & work on enabling local deployments for sprint next week 16:07 (finito) 16:08 pitti: persistent net names, had we not fixed that already with the previous trusty dot release? 16:08 doko: hehe, only for tomorrow? ;-) 16:08 tested connection to new DSE servers (good) 16:08 discussion with stub re tables from old cassandra dbs to import 16:08 release upgrade testing from Trusty to Vivid 16:08 searched for / consolidated duplicates of T to V upgrade failures (LP: #1534374) 16:08 Launchpad bug 1534374 in ubuntu-release-upgrader (Ubuntu) "unable to upgrade to 15.04 due to libstdc++6 SRU" [High,Triaged] https://launchpad.net/bugs/1534374 16:08 cyphermox: apparently not 16:08 investigation into u-r-u saying packages are unauthenticated 16:08 release upgrade testing from Trusty to Wily 16:08 reported LP: #1537900 re dist-upgrade failure to wily 16:08 uploaded sysvinit fix for LP: #1507151 to wily-proposed 16:08 Launchpad bug 1507151 in sysvinit (Ubuntu Wily) "duplicate for #1537900 sysv-rc.postinst calls insserv by name, but insserv package does not provide the command in a bin directory" [High,Fix committed] https://launchpad.net/bugs/1507151 16:08 updated ubuntu-release-upgrader to support Trusty to Wily upgrades 16:08 Launchpad bug 1507151 in sysvinit (Ubuntu Wily) "sysv-rc.postinst calls insserv by name, but insserv package does not provide the command in a bin directory" [High,Fix committed] https://launchpad.net/bugs/1507151 16:08 W SRU verification of LP: #1537916, #1507151 16:08 worked on britney sending email notifications 16:08 patch piloting 16:08 SRU queue processing 16:08 Launchpad bug 1537916 in ubuntu-release-upgrader (Ubuntu Wily) "wily release upgrader needs to support upgrades from Trusty" [High,Fix committed] https://launchpad.net/bugs/1537916 16:08 installed / setup Ubuntu on new laptop 16:08 pitti: (we removed biosdevname in 14.04.3 for much of the same thing) 16:08 ✔ done 16:08 tdaitx, well it's a special fosdem beer location, with extra special beers and tokens. 16:08 pitti: speaking of armhf and scalingstack, did you happen to see the post from hrw a few weeks ago about "how I got 32-bit arm VMs running right"? 16:09 cyphermox: err, no, we didn't -- biosdevname is still used in d-i in latest netboot 16:09 slangasek: uh, no, I didn't.. showmeshowmeshowme! 16:09 maybe I'm mistaking it for some other release then 16:09 slangasek: I was talking to wgrant about that and made some experiments, but there were still some hiccups 16:09 pitti: https://marcin.juszkiewicz.com.pl/2016/01/17/running-32-bit-arm-virtual-machine-on-aarch64-hardware/ 16:09 bdmurray, any proposed action on 1534374? 16:09 this is on Fedora, but hopefully translates 16:10 slangasek: oh, that way around -- I tohught you meant "boot armhf images in scalingstack" 16:10 bug 1534374? 16:10 bug 1534374 16:10 bug 1534374 in ubuntu-release-upgrader (Ubuntu) "unable to upgrade to 15.04 due to libstdc++6 SRU" [High,Triaged] https://launchpad.net/bugs/1534374 16:10 doko: wait for Vivid to End of Life 16:10 pitti: well, but it should be bootable in scaling stack if we get the pieces figured out :) 16:10 bdmurray: what is "DSE", OOI? 16:10 bdmurray, I like that 16:10 pitti: data stack enterprise 16:11 will vivid actually EOL? I thought touch releases are still based on vivid 16:11 pitti: cassandra w/ support and bells and whistles 16:11 Oh crap 16:12 * uplift edk2 to new upstream snapshot per request of hyperscale team 16:12 * ppc64el triage for 14.04.4 16:12 * merges: at a personal 5-year low for outstanding TIL merges ;) 16:12 * nudging packages for proposed-migration 16:12 * MP review for system-image "device alias" support 16:12 * forward progress on P7 machine deprecation 16:12 * discussion about Archive Reorg changes for 16.04 (headline: build-depends won't have to go through MIR, only binary depends) 16:12 * sil2100 is obviously very late but still needs time to prepare his report 16:12 sil2100, are you not looking forward to vivid builders getting turned off? =) 16:12 (done) 16:12 LP #1535349. Patch created waiting on sponsorship. Discovered an additional issue with initramfs-tools that will need to be resolved before this is fully fixed. 16:12 LP #1484696. Fix Released issue closed. 16:12 Xenial issues with pulseaudio rejecting connections in xenial. Daemon is still up, but not responding. Case to be found or reported. 16:12 Xenial issues where chrome does not respond to dbus url open requests. 16:12 --done-- 16:12 Launchpad bug 1535349 in coreutils (Ubuntu Trusty) "`df /dev/sda1` no longer reports information for /dev/sda1" [Medium,Confirmed] https://launchpad.net/bugs/1535349 16:12 slangasek: you can actually create an AMI with the kernel/initrd and boot the armhf cloud image -- but that fails on some libvirt issue (which wgrant already dealt with by backporting the fix) and some scalingstack config issue 16:12 Launchpad bug 1484696 in MAAS 1.10 "Unable to connect to: ws://:/MAAS/ws" [Undecided,New] https://launchpad.net/bugs/1484696 16:12 sil2100: aim to go right after infinity, he is last on the list ;-) 16:13 - binutils 2.26 release 16:13 - gcc-5 ibm backports 16:13 - icedtea-web update, build for openjdk-8 16:13 - fix firefox build on arm64 16:13 - getting gcc cross packages mostly installable 16:13 - ruby2.2 update 16:13 - openjdk-6, openjdk-9 uploads 16:13 tdaitx: thanks 16:13 - work on python3.4 removal 16:13 - work on ruby2.1 removal 16:13 - some work on multiarchifying packages 16:13 slangasek: oh? where's that discussion happening? 16:13 (done) 16:13 slangasek: uh, no "main" closure on b-deps any more? that sounds a bit strange 16:13 barry, slowly being worked on ;-) 16:14 barry: it's been internal discussion up to this point, I'll post on ubuntu-devel shortly about it 16:14 just curious if there's a public thread on it? 16:14 oh cool 16:14 pitti, it's not as weird as it sounds. as vastly smaller in scope. 16:14 it's in the same general vein as Archive Reorg has always been - "security support over the binary dep closure, not over the source dep closure" 16:15 xnox: "Go static linking" 16:15 pitti, and C++ templates 16:15 we just found a shortcut to the implementation 16:15 pitti, are two things to watch out for, yes. 16:15 pitti, xnox: Built-Using 16:15 slangasek, oh, ok. 16:15 i.e. we should still be closed over Built-Using, just not over Build-Depends 16:15 so if we render main packages unbuildable because of some universe crap nobody cares about, we make our lifes even harder 16:16 xnox: your turn 16:16 pitti: happy to make this a discussion topic later in the meeting :) 16:16 Does anybody know if it's possible to start an lxc from inside a schroot? 16:16 robru, we are in a meeting =) 16:16 working on fixing cloud images (borked link_in_boot) 16:16 working on publishing updated installer guide 16:16 fixed .ins and el torito boot files 16:16 packaged libica 16:16 upgraded btrfs tools, to fix bugs on ppc64el 16:16 attended openmainframe project eu meetup in luton 16:16 off to fosdem tomorrow 16:16 etc. 16:16 done 16:16 robru: a chroot by itself has no namespacing; so yes it should be possible as long as you have all the right mounts in the chroot 16:17 sla claws-mail: merged w/debian, uploaded 3.13.1-1.1ubuntu1 and 3.13.1-1.1ubuntu2 (the latter to fix upstream bug #3600) 16:17 bug 3600 in Launchpad itself "Summary field processing is handling carriage returns wrongly." [Medium,Fix released] https://launchpad.net/bugs/3600 16:17 dirtbike/pip stack: everything working in my local staging environment. now i need to 1) release dirtbike & upload to debian; 2) upload python-progress (debian bug #812908); 3) update debian python policy for wheels; 4) update pip to 8 & upload. it was lots of work, but it's all paying off now and will make ongoing maintenance *much* easier, and will finally get us a much more modern pip. next up: virtualenv 16:17 Debian bug 812908 in wnpp "ITP: python-progress -- easy progress reporting for Python" [Wishlist,Open] http://bugs.debian.org/812908 16:17 worked on the foundations-x-python3-only blueprint. samba-libs is still the big blocker. reached out to the fedora porter who has made some progress, but still a lot of work to do. this one has me the most worried. 16:17 i won't wait for the debian gnome maintainers much longer on libpeas. if nothing happens in debian by the end of next week's sprint, i'll move forward in ubuntu. 16:17 --done-- 16:18 sil2100: no infinity, go 16:18 o/ 16:18 barry, is there an open issue? 16:18 - Landing team work, silo coordination, preparing landing e-mails 16:18 - RTM Status meetings 16:18 - system-image: 16:18 * Coordinating the removal of manta images from official s-i 16:18 * Poke community s-i hosts to pick up manta once we remove those 16:18 * Tests tests 16:18 doko: on libpeas? 16:18 - Promoting devel images 16:18 yes 16:18 - OTA-9: 16:18 * Preparing release notes 16:18 * Releasing the images (with some complications due to management miscommunication) 16:18 * Investigating various reports 16:18 - Future OTA schedule preparation, announcements 16:18 - Multiple secret work involving device enablement in s-i 16:18 doko: yes, let me find it 16:18 - Preparing a new merge for ppp, symbols changes needed 16:18 - Help in preparing silo 12 landing, packaging reviews etc. (a big PD-related silo) 16:18 - OTA-9.5 16:18 * Planning and scheduling, various long discussions 16:18 * Add new frameworks to the seeds and store 16:19 - Landing-team-tools - write a helper script for getting s-i image information 16:19 (done) 16:19 slangasek: what are the right mounts? It has /proc and /dev/pts 16:19 sil2100: congrats on OTA-9 16:19 Thanks, some people do seem to have issues with seeing the update though, will have to look into that a bit 16:19 Since the phasing ended like 6-8 hours ago 16:19 slangasek: so yes, that's using direct kernel boot much like I attempted on scalingstack; this should work in principle indeed 16:20 But yeah, in overall it's good :) 16:20 doko: LP: #1440504 which has a link to the debian bug 16:20 Launchpad bug 1440504 in libpeas (Ubuntu) "libpeas-1.0-0 depends on both libpython2.7 and libpython3.4" [Medium,In progress] https://launchpad.net/bugs/1440504 16:20 doko: debian bug has patches 16:20 robru: I don't know all of them, but there'll be at least /sys and I think there's at least one other mount needed for the cgroups 16:20 (but I have no cgroups fs mount on my system currently, hmmm) 16:20 maybe that's old info 16:21 o/ 16:21 sorry I'm late 16:21 pitti: ah, direct kernel boot; that's obviously less than perfect, but I guess it's the only option we have there 16:21 slangasek: tmpfs on /sys/fs/cgroup ? 16:21 slangasek: it's what the blog uses too 16:21 pitti: oh haha I was looking at mounts inside a chroot, so there 16:22 caribou: hi! anything you'd like to report? 16:22 yep 16:22 LP1522346 - nut merge 16:22 LP1536904 - kdump fails on 16.04 16:22 LP1537714 - smaller initrd for older kernels 16:22 Sponsorship : LP1089013 - clvm 16:22 Sponsorship : LP1248054 - dlm 16:22 and rework of my home network 16:22 (done)à 16:22 caribou: thanks for nut, I'm glad to no longer be TIL on it ;-) 16:23 now that just leaves me shadow 16:23 oh, & looks like there will be another rsyslog merge needed 16:24 I can also give a status update for infinity 16:24 slangasek: so is it enough to just bindmount from the host into the chroot? Where's the documentation on everything lxc needs? 16:24 * working on glibc 2.22 merge and reconciling locales packages with Debian 16:24 slangasek: I mean to merge shadow 16:24 robru: I don't know; possibly somewhere near the lxc package 16:25 cyphermox: oh? but I'm TIL :) 16:25 robru, try stgraber ? 16:25 =) 16:25 slangasek: ah, I can leave it to you then ;) 16:25 cyphermox: maybe you want to merge console-setup instead, you already have your name down on that one on merges.u.c :) 16:25 ok, any questions on status? 16:26 slangasek: yeah, that one too 16:26 [TOPIC] Archive Reorg 16:27 since there were questions about this, let's discuss 16:27 pitti: you had concerns about uncared-for packages in universe 16:28 cyphermox, ohh, I started shadow, but lost track. while the debian maintainer took most patches, they are disabled, and need an update 16:28 slangasek: well, yes; mostly about rendering more main packages FTBFS and thus making it harder to remove obsolete universe packages 16:28 doko: ok.. I mentioned it because you asked me about a shadow upload to trusty :P 16:29 but also with all the static linking madness of go, where build deps are really binary deps 16:29 pitti: so the reality today is that we spend a lot of effort going through the MIR process for these same packages, pulling them into main 16:29 and then we still don't do anything with them because they're not actually interesting, they're only needed as build-dependencies 16:30 I expect this to be a net reduction in work effort, because a) less MIR paperwork for things we don't care about and don't want to support, b) less packaging delta from Debian to enforce a build-time main/universe split, c) fewer packages in main overall meaning less security team work 16:30 pitti, my thoughts were that packages that main build-depends on, will become known as "supported-build-depends" hosted mostly in universe. And then we can implement controls and/or additional policy on that. For example, uploads only by core-dev, rather than motu. 16:30 this btw was always part of the ArchiveReorg plan, which we first discussed back in Barcelona :) It only foundered on some of the implementation details 16:31 pitti, (where supported-build-depends is a seed) 16:31 slangasek: oh yeah, that's the other thing -- we have some packages where we drop build deps to drop functionality which is only available with universe b-deps, for building plugins for example 16:31 xnox: fwiw I don't see any reason to restrict uploads to core-dev 16:31 slangasek, ok. 16:31 slangasek: you'd have the same delta if they end up being binary deps (new library), or an invisible "static linking" case when they don't become build-deps 16:31 pitti: static linking of go MUST be tracked already via the Built-Using field, which I mentioned above 16:31 pitti, yeah. and e.g. especially when those plugins, will be published into universe anyway. We already do build a lot packages in main, that have some binary packages in universe. 16:31 cyphermox, feel free to take over 16:32 hmmm I wonder if this would actually cause python2 to drop out of main in 16.04 ;) 16:32 doko: *shrugs* I will finish console-setup first, I already have it in a PPA ready to finish testing 16:33 slangasek, i wonder if "built-using" should be included in main closure then (whereas "build-depends" will not) 16:33 slangasek: wouldn't that be wonderful? :) 16:33 slangasek, ubuntu-desktop depends on python2 at the moment. 16:33 pitti: the forcing function for 16.04 is nodejs. This is something the security team will *not* be providing support for, yet it's pulled in as a build-dependency for documentation 16:33 xnox: sorry, that's exactly what I meant 16:33 slangasek, ack. 16:34 xnox: see my previous comments about samba-libs :( 16:34 xnox: Built-Using should be included in main (and therefore component-mismatches should report on it), Build-Depends should not 16:34 pitti: but there are other examples besides nodejs; maven has been one this cycle as well, with its many tentacles of java 16:35 yes, if the build-dependency translates to a binary dependency, we still have to deal with it 16:35 * barry will be happy to drop some debian deltas because of this 16:35 but the reason for ArchiveReorg is the recognition that there are many packages that are in main only because they're build-dependencies and *not* runtime dependencies (or statically-linked) 16:35 well, I can't say I have a good feeling about this, but maybe that's just me then 16:37 pitti: it'll reduce the workload, trust me ;-) 16:37 slangasek: oh, no doubts about that, I'm more afraid of piling up hidden traps 16:37 hmm, well 16:38 the kind of trap you've mentioned - bitrotting package in universe - is still something we'd be responsible for sorting out, in either case 16:38 slangasek, xnox it would be nice to have some high level archive spelunking tooling around this. e.g. given a package, tell me the status of all its b-d and depends. or going the other way, given a package, tell me the status of its reverses 16:38 barry, that would be check-mir =) 16:38 i.e. if it's a build-dep, we're responsible for sorting out the bitrot; it doesn't matter if that package is in main like today, or if we move it to unvierse 16:38 barry, we are still keeping main/universe components 16:39 xnox: well, we need to stand up a modified component-mismatches and germinate 16:39 slangasek: right, but so far we've just dropped the build-dep of "meh, don't like that" stuff beforehand, instead of letting them creep in everywhere 16:39 xnox: something like that, probably with some enhancements 16:39 * xnox really should share my draft document with everybody. 16:39 so this will undoubtedly work well initially; my fear is that it will make it harder down the road, as we effectively end up maintaining half of universe 16:40 maven, haskell, what not 16:40 but we already do 16:40 I disagree that this means we will be maintaining maven or haskell 16:41 e.g. haskell transitions, java transtions it's still we do anyway. 16:41 except to a very basic level 16:41 a healthy main, requires a somewhat healthy universe. 16:41 * xnox was writting down my thoughts on the subject in a google doc -> https://docs.google.com/document/d/1dJBtLLCppH2yt664S8G2jB_tK-iWi_D7wqaN6S4ddwI/edit# 16:42 xnox: have you checked past archive reorg documentation in the wiki already? 16:42 slangasek, i've read and re-read most of it. And it is confusing. 16:42 \o/ 16:43 slangasek, cause parts were implemented, and other parts are not (e.g. we did permissions, but we didn't abolish components) 16:43 slangasek, and this new "supported-build-depends" subject, i'd rather not call "archive-reorg", as the scope is smaller. 16:43 e.g. we are keeping M.I.R. and we are keeping a show-case pinacle "main" component. 16:44 we were always going to be keeping an MIR process 16:44 and kicking things out of it, that are not pinacle show-case stuff =) 16:44 it was just a question of whether the includer was still called "main" 16:44 slangasek, well yes. But the whole apt preferences shananigans were nuts. 16:44 heh 16:45 my plan was to generate sample new seed output, sample new components missmatches, polish the document 16:46 and like seek further feedback 16:46 but i guess the cat is out of the bag now. 16:46 don't worry, you'll get all the feedback you need ;) 16:46 [TOPIC] AOB 16:46 anything else today? 16:47 I have a small proposal to replace the Linux kernel in main with Hurd 16:47 *hurt 16:47 slangasek: pff 16:47 I'd rather you pick Darwin. 16:47 heh 16:47 ;) 16:48 nah, keep it & rewrite in Go 16:48 slangasek, +1 on Darwin 16:48 that'll dovetail nicely with my plans to rewrite go in guile 16:48 stuff I didn't have time to write: I will be off tomorrow and from Wed to Frid next week (swap day + vacation), and then the week after on Monday and Tuesday (holidays); moved from hexchat to quassel so now I can have it connected most of the time (just in time for some time off) 16:48 barry, using an llvm compiler right? 16:48 Mir 3.0, now with DisplayPostscript support 16:49 xnox: of course! 16:49 slangasek, \o/ 16:49 slangasek, requires a type-c dongle? 16:49 xnox: DisplayPostscript is over CP only 16:49 tdaitx: oh good, enjoy the scrollback then ;) 16:50 cyphermox, i have no idea what CP is - is that available on a mainframe? 16:50 yeah, as if the email backlog is not enough ;-) 16:50 xnox: I suppose. Carrier Pidgeon? 16:50 * barry fires up his old NeWS server 16:50 #endmeeting