== Meeting information == * #ubuntu-meeting Meeting, 02 Oct at 15:03 — 15:37 UTC * Full logs at [[http://ubottu.com/meetingology/logs/ubuntu-meeting/2014/ubuntu-meeting.2014-10-02-15.03.log.html]] == Meeting summary == === Lightning round === The discussion about "Lightning round" started at 15:03. === AOB === The discussion about "AOB" started at 15:19. === kernel crash dumps === The discussion about "kernel crash dumps" started at 15:21. == Vote results == == Done items == * (none) == People present (lines said) == * mvo (64) * caribou (29) * barry (17) * slangasek (16) * sil2100 (16) * bdmurray (16) * robru (11) * ubottu (11) * jodh (10) * stgraber (8) * doko (8) * meetingology (4) == Full Log == 15:03 #startmeeting 15:03 Meeting started Thu Oct 2 15:03:25 2014 UTC. The chair is slangasek. Information about MeetBot at http://wiki.ubuntu.com/meetingology. 15:03 15:03 Available commands: action commands idea info link nick 15:03 hi 15:03 o/ 15:03 [TOPIC] Lightning round 15:03 $ echo $(shuf -e barry doko stgraber jodh bdmurray slangasek cjwatson caribou infinity mvo bhuey sil2100 robru) 15:03 mvo infinity sil2100 slangasek robru stgraber jodh caribou bdmurray barry doko bhuey cjwatson 15:03 first! 15:04 apt: 15:04 - Diagose failure that Daniel Westervelt had with armhf/maas 15:04 (bug in Acquire::GzipIndexes code but already fixe) 15:04 - Enable all hardening options (including PIE) 15:04 - Smoe more work on acq-trans branch 15:04 click: 15:04 - Deal with libqt5gui5 -> -gles fallout (emulator releated) 15:04 - Fix gnutls error for public.apps.u.c in their deploy charm 15:04 (lp:~mvo/canonical-is-charms/ubuntuone-servers-deploy-public.apps.u.c) 15:04 - Address review comments for lp:~mvo/click/acquire 15:04 - Port acquire to use pythons new asyncio (lp:~mvo/click/acquire-asyncio) 15:04 - Port lp:~mvo/click/sso to pycurl and resolve gnutls issues (missing 15:04 ServerName in server config for TLS) 15:04 - Start discussion with SDK team about lp:~mvo/click/click-check-libs 15:04 - Work on fixing lp:~mvo/click/lp1232130-kill-on-remove-2 15:04 (tricky because ubuntu-app-launch is on the session bus but click is root) 15:04 - Port lp:~mvo/click/repository to use pycurl and provide httplib2 like 15:04 http_request() function to make porting simpler 15:04 - Review lp:~cjwatson/click/chroot-name 15:04 - Update lp:~mvo/click/lp1319790-chroot-fstab 15:04 lp:~mvo/click/dont-crash-for-empty-db, 15:04 lp:~mvo/click/lp1232130-kill-on-remove, 15:04 lp:~mvo/click/lp1219912-build-exclude, 15:04 lp:~mvo/click/framework-info to current click 15:04 click-bin-path: 15:04 - Debug/fix crash when nothing has a bin-path 15:04 sponsoring: 15:05 - Review/upload ubuntu-core-config/initramfs-tools-ubuntu-core 15:05 system-image: 15:05 - Upload pycurl based image downloader to the foundations PPA 15:05 - Debug build failure with pycurl in system-image ppa build 15:05 - Debug/fix incorrect writes with lambda and write custom writer 15:05 - Look into unneeded core image dependencies 15:05 - Port pycurl branch to CurlMulti 15:05 - make total progress totally reliable 15:05 - add get_download_manager() factory 15:05 - update default config for ppa system-image-cli 15:05 - tweak seeds, do not include recommends in system-image seed 15:05 (done) 15:05 mvo: \o/ 15:06 no infinity at the moment 15:06 Ah, ok 15:06 sil2100: 15:06 - Landing team work, preparing landing e-mails 15:06 - CI Train maintenance and feature: 15:06 * Deployment of the build job refactoring 15:06 * Refactoring of the publish job 15:06 * Adding unit tests to the few newly ripped out publish job chunks 15:06 * Work on the dual landing scheme for ubuntu-rtm 15:06 * Start looking into the issue of not all bug numbers landing in debian/changelog 15:06 - Debugging and fixing problems in commitlog generation (google API problems) 15:06 - Sending out announcements about changes in some of our processes 15:06 - Packaging and quilt advice for some upstream projects 15:06 - Preparing script for ubuntu-rtm backlog generation 15:06 - Health problems, one day of sickness, slow recovery - less work done... 15:06 barry: get_download_manager() should actually be good now, it will check the bus for udm and use that if its available and otherwise try to fallback to pycurl 15:06 (done) 15:07 mvo: that's really great 15:07 mvo: i might still want the setting to override autodetection, but we'll see 15:07 barry: thats added as well :P 15:08 barry: via the environment right now though for easy testing 15:08 mvo: yep, makes sense 15:09 * feeling ill this morning, so not a very comprehensive report 15:09 * partner package reviews 15:09 * working on various branch reviews (upstart, apport) 15:09 (done) 15:09 robru: 15:09 barry: whats the best irc channel to ask about asyncio for "normal" files? and if its worthwhile to work on a patch? 15:10 #chair mvo 15:10 Current chairs: mvo slangasek 15:10 mvo: good question. *maybe* #python-dev? i have to see if there's a specific asyncio channel 15:10 (as I may not make it to the end of the meeting today) 15:10 looks like no robru 15:10 stgraber: 15:10 ah sorry, need a minute 15:10 robru looked alive ;) 15:11 oops, sorry 15:11 been preparing for LinuxCon and Plumbers, did the LXC 1.0.6 release, did the LXC 1.1~alpha2 release, got a FFe for that latter and uploaded to utopic 15:11 ok, robru after stgraber then 15:11 now trying to track down a very annoying kernel bug that's affecting my server 15:12 mvo: https://code.google.com/p/tulip/ (tulip being the code name for the standalone version) 15:12 I'll be gone next week except for Friday, then I'll be gone the week after at LinuxCon/Plumbers and the week after that for the device sprint. 15:12 fully back on the 27th 15:12 (DONE) 15:12 - ripped all knowledge of spreadsheets out of the CI Train jenkins backend. It is now 100% ready to have something else replace the spreadsheet (just waiting for the spreadsheet replacement itself to be ready). 15:12 - work continues on encapsulating the various bits of spaghetti that exist in the ci train code. eliminated knowledge of the json configuration format from the prepare job, check-publication-migration job, and the build job. 15:12 - made it easier to rollback production deployments in case I land something broken 15:12 - resurrect accidentally deleted copy2distro script and brought it up to current coding standards (pep8, pyflakes, etc) 15:12 - fixed all kinds of little bugs and typos and stuff 15:12 - ported most of the bash scripts to sh, except for one in which sh breaks pbuilder for some unknown reason. 15:12 - eliminated an entire class of ridiculous path-traversing code which was hunting for silos... which exist only in a single well-defined location. 15:12 - various and sundry landings. 15:12 - vacation next week! 15:12 (done) 15:12 * upstart 15:12 - bug 1198180 investigations with apw. 15:12 bug 1198180 in upstart (Ubuntu) "possible leak in upstart 1.5" [Medium,Confirmed] https://launchpad.net/bugs/1198180 15:12 * goget-ubuntu-touch: 15:12 - Raised MP to add '--list-images' option: 15:12 https://code.launchpad.net/~jamesodhunt/goget-ubuntu-touch/add-list-images-option/+merge/236510 15:13 * system-image 15:13 - Lots of activity. Currently investigating improvements to upgrade 15:13 logic and will then start serious package whittling. 15:13 lxc autopkg tests fail 15:13 🀫 15:13 doko: that's normal 15:13 jodh: nice work on the goget branch 15:13 ahh 15:14 mvo: w/e hack :) 15:14 doko: see part about the kernel panic on my server :) adt will pass again once the public images server works again 15:14 * sosreport 3.2 released 15:14 - Started to work on Debian/Ubuntu packaging 15:14 - Continue implementation of git-dpm 15:14 - Found zero-day bug in apt plugin 15:14 * Investigation on a iSCSI/LVM/DM-MPIO mount issue at boot 15:14 * Preliminary work on running kdump-tools at RunLevel S 15:14 (done) 15:15 caribou: git-dpm ftw! 15:15 barry: looking forward to your talk != 15:15 :) 15:15 modified errors charm and code to create version_info 15:15 research into volume of CORE files being reported to the Error Tracker 15:15 searched for CORE files to be removed from swift 15:15 r537: daisy/submit.py: do not ask for a CORE for packages without Ubuntu or Ubuntu RTM as we will not be able to retrace them 15:15 r538 "daisy/submit.py: add in a metric for retraced crashes missing a stacktrace" 15:15 r540: daisy/submit.py: do not ask for CORE files if the report is missing information required for retracing the crash" 15:15 reworked steve's apport changes for whoopsie-upload-all 15:15 investigation into ubuntu-touch default test whoopsie failure and discussion with plars about it 15:15 reviewed paul larson's change to whoopsie default ubuntu-touch test (approve) 15:15 submitted apport bug LP: #1374544 regarding SegvAnalysis field weirdness 15:15 reported apport bug LP: #1376374 regarding repeat processing of crash files 15:15 Launchpad bug 1374544 in apport (Ubuntu) "crash file indicates Disassembly is missing when it isn't" [Undecided,New] https://launchpad.net/bugs/1374544 15:15 Launchpad bug 1376374 in apport (Ubuntu) "whoopsie-upload-all will run hooks on a corrupt crash file multiple times" [Undecided,New] https://launchpad.net/bugs/1376374 15:15 SRU verification of bug LP: #1324833 15:15 Launchpad bug 1324833 in aptdaemon (Ubuntu Trusty) "crash handler does not include package version" [High,Fix released] https://launchpad.net/bugs/1324833 15:15 uploaded fix for update-manager LP: #1374715 regarding CVE urls 15:15 Launchpad bug 1374715 in update-manager (Ubuntu) "CVE links in the updater are invalid" [Medium,Fix released] https://launchpad.net/bugs/1374715 15:16 uploaded trusty SRU for LP: #1354571 so we'll stop uploading crashes with bad core dumps to the error tracker 15:16 Launchpad bug 1354571 in apport (Ubuntu Trusty) "apport-retrace ignores warnings from gdb" [Medium,Fix committed] https://launchpad.net/bugs/1354571 15:16 submitted merge proposal fixing sru-report traceback 15:16 ✔ done 15:16 system-image: LP: #1370586; LP: #1376758 (triaged); system-image 2.5-0ubuntu1 (monkeypushing rebuilds, citrain, rtm, testing). reviewed mvo branch. 15:16 Launchpad bug 1370586 in Ubuntu system image "Add synchronous method to determine if there are known updates" [High,Fix released] https://launchpad.net/bugs/1370586 15:16 Launchpad bug 1376758 in Ubuntu system image "Ability to specify script to call at end of download rather than rebooting" [High,Triaged] https://launchpad.net/bugs/1376758 15:16 debuntu: tox 1.8.0-1. python-concurrent.futures 2.2.0-1. tons of pycurl bug triaging, conversion to git (for better debian/ubuntu package management), 7.19.3-1-2, 7.19.5-1, 7.19.5-2, and LP: #1376736 (FFE) with 7.19.5-2ubuntu1 ready to go. debian bug #763770. debian sbuild creation (fails due to debian bug #607228) 15:16 Launchpad bug 1376736 in pycurl (Ubuntu) "[FFe] update to pycurl 7.19.5" [Undecided,New] https://launchpad.net/bugs/1376736 15:16 Debian bug 763770 in src:pycurl "pycurl: Add DEP-8 tests" [Normal,Open] http://bugs.debian.org/763770 15:16 Debian bug 607228 in sbuild "no way to run setup command inside a chroot" [Normal,Fixed] http://bugs.debian.org/607228 15:16 other: reviewed branches for cgoldberg and robru. falcon upstream branch rebase and back/forth to get landed. 15:16 more git-dpm hacking 15:16 -- done -- 15:17 doko is next 15:17 - vacation last week 15:17 - catching up with email 15:17 - toolchain updates 15:17 - python update 15:17 - a bit of MIR processing 15:17 (done) 15:17 and now bhuey 15:18 and cjwatson is not here today 15:19 no bhuey? 15:19 [TOPIC] AOB 15:19 anything else? anything interessting happening last week thats worth sharing? 15:19 I have: asyncio is cool 15:21 nothing…? 15:21 mvo: continue! :) 15:21 [TOPIC] kernel crash dumps 15:21 caribou: do you want to talk a little bit about this? 15:22 mvo: yes, it's *super* cool 15:22 barry: heh :) everybody want to get back to real work eh ? 15:22 mvo: maybe a brief update if someone is interested 15:22 caribou: yes! 15:23 the Debian/Ubuntu way of capturing a kernel crash dump is, once kexec triggers a reboot following a kernel panic 15:24 to run a SysV initscript called kdump-tools that will do the necessary steps to read from /proc/vmcore 15:24 use makedumpfile to process the /proc/vmcore output and to some filtering and compression and to produce a file 15:25 in /var/crash/{timestamp dir} directory. the content of the dmesg output is also captured in that directory 15:26 the system is then rebooted. On startup Apport will produce a Report file in /var/crash that contains the output of the dmesg file 15:26 that Apport report used to contain the full kernel crash dump but given the size of those files, it is now left in the timestamped directory 15:27 the kernel crash file can be later analyzed in conjunction with a kernel namelist which includes the debug symbols 15:28 I recently introduced in the latest version available in Debian (1.5.7) a functionality to allow the crashdump file to be sent to a remote server via SSH or NFS 15:29 There is also some preliminary work being done to have the kernel dump capture to be executed at run level S 15:29 caribou: is there sensitive information in the crash dump, and if so, does it get trimmed out before it leaves the local machine? 15:30 there can be, especially if all userspace memory is capture (the default is not to capture them) 15:30 but network information, in-transit data & such is bound to be present in the crash dump file 15:31 which is why many user are very cautious of what is being done with those files. Some even refuse to make them 15:31 available, so sometime the analysis has to be done remotely on their servers 15:32 For instance, the file system structure, file names, TCP/IP information & adresses, process list with arguments, all of those are easily available with the default setup 15:33 It is makedumpfile responsability to filter out unneeded/unwanted memory pages 15:34 that's about all I have if I want to remain brief 15:34 cool, interessting stuff, thanks for this! 15:34 oh, for some reason it seems much more difficult to get a crash dump on a laptop/desktop 15:34 and next week we will have cjwatson talking about (one of the things) he is working on 15:34 I have an outstanding bug on that 15:35 oh, what the issue here? 15:35 caribou: thanks! 15:35 yw 15:36 more comments/questions? if not it seems we are close to EOM :) 15:37 #endmeething 15:37 thanks everyone! 15:37 #endmeeting even Generated by MeetBot 0.1.5 (http://wiki.ubuntu.com/meetingology)