[00:02] <wgrant> timrc: Hi.
[00:03] <wgrant> Ah, I se.
[00:36] <lifeless> StevenK: I know why parallel is hanging
[00:36] <lifeless> StevenK: contention on template1 copies.
[00:40] <lifeless> StevenK: and on launchpad_ftest_template, I have a patch to sort that, but I'm getting the template1 (lower level tests) sorted first
[00:40] <lifeless> OperationalError: source database "template1" is being accessed by other users
[00:40] <lifeless> DETAIL:  There are 6 other session(s) using the database.
[00:42] <lifeless> hmm, 30 seconds to do a full schema setup including buildout wadl etc
[00:42] <lifeless> plausible to make it inline on layer setup
[00:42] <wgrant> lifeless: scan_branches seems to be hung.
[00:43] <wgrant> Can you look at it, please?
[00:43] <lifeless> oh joy
[00:43] <lifeless> thanks
[00:43] <LPCIBot> Project devel build #851: FAILURE in 5 hr 0 min: https://lpci.wedontsleep.org/job/devel/851/
[02:47] <james_w> lifeless, hi, how do I get subunit-filter or something to not print the "time:" lines?
[03:21] <lifeless> james_w: hi
[03:21] <lifeless> james_w: the time: markers will be consumed by whatever human UI you use
[03:22] <lifeless> james_w: e.g. subunit2junixml, subunit2pyunit, tribunal, subunit-stats
[03:22] <james_w> my eyes don't consume it :-)
[03:22] <lifeless> if you're seeing hundreds of adjacent time: statements, get a new python-subunit, it now coalesces them
[03:22] <james_w> I was using subunit-filter to just get me the skipped test so I could see what it was
[03:23] <lifeless> yeah
[03:23] <james_w> adding subunit2pyunit helped, thanks
[03:23] <lifeless> thats the case where you used to see many time statements together
[03:23] <lifeless> subunit-ls may be useful for you too
[03:23] <lifeless> anyhow, that case is fixed in newer subunits
[03:23] <james_w> subunit-ls: AttributeError: 'NoneType' object has no attribute 'id'
[03:23] <james_w> so not really :-)
[03:23] <lifeless> as is that :)
[03:23] <james_w> yeah
[03:25] <james_w> 0.0.7 works much better, thanks
[03:30] <lifeless> wheee
[03:30] <lifeless>  306 /   36  Distribution:EntryResource:searchTasks
[03:30] <lifeless>      154 /   37  Distribution:+bugs
[03:30] <lifeless> need to do someting about that
[04:07] <lifeless> well
[04:07] <lifeless> the world must be ending - ms shipping gcc
[05:13] <lifeless> wow, postgresql, creating a db isn't really tuned is it
[05:14] <mwhudson> lifeless: someone did an experiment years ago where lp untarred a db rather than creating one with pg; didn't save any time
[05:16] <lifeless> mwhudson: bin/test --parallel -vvt canonical.database.postgresql
[05:17] <lifeless> you'll want the bugfix from my paralleltests branch
[05:17] <lifeless> or they will try to create the *same* test db
[05:18] <lifeless> mwhudson: we're seeing 10 second delays sometimes
[05:18] <lifeless> with multiple threads making db's
[05:19] <mwhudson> lifeless: ah so it's kinda overdone locking or something?
[05:20] <lifeless> underdone it seems
[05:20] <lifeless> createdb works by cloning (e.g. template1)
[05:20] <lifeless> if two users try to clone at once, the *both* start, and then *both* fail ('other users are accessing template 1')
[05:21] <mwhudson> oh
[05:21] <mwhudson> nice
[05:21] <lifeless> s/the/they/
[05:21] <lifeless> 200ms to clone template1 on my new machine
[05:21] <lifeless> 8 hardware threads
[05:21] <lifeless> how often do you think I hit this contention ?
[05:25] <lifeless> being able to set a connection limit on the db helps
[05:27] <lifeless> but it varies wildly
[05:27] <lifeless> even with a bzr lock around our create db calls:
[05:27] <lifeless> 1309494289.41 (12.18) completed (1 retries) 4796 template1
[05:27] <lifeless> released 4796
[05:51] <LPCIBot> Yippie, build fixed!
[05:51] <LPCIBot> Project devel build #852: FIXED in 5 hr 7 min: https://lpci.wedontsleep.org/job/devel/852/
[07:13] <lifeless> wgrant: did you end up getting lxc rabbit fixture happy?
[07:37] <wgrant> lifeless: No. Need to investigate that on mondayish.
[08:03] <lifeless> I have bad news about non-lxc parallel tests
[08:10] <lifeless> 311 seconds (serial) -> 92 (parallel) with my 8-core, and I think its all contention on the db reset. I need to profile more.
[08:11] <lifeless> bzr+ssh://bazaar.launchpad.net/~lifeless/launchpad/paralleltests has my stuff so far
[08:28] <wgrant> lifeless: Is that with multiple templates?
[08:28] <wgrant> lifeless: Have you tried multiple clusters?
[08:42] <lifeless> wgrant: thats with a double-template
[08:42] <lifeless> wgrant: each runner forks l_f_t to l_f_t_$pid
[08:42] <lifeless> wgrant: then makes teset instances from the l-f_t_$pid
[08:47] <lifeless> wgrant: I'm going to do family time now, but perhaps you'd like to review https://code.launchpad.net/~lifeless/launchpad/paralleltests/+merge/66554
[10:03] <wgrant> rvba: You'll need to add 'job.notifications-queue.' to the subscribe_key when binding.
[10:03] <wgrant> Well, 'job' is at your discretion.
[10:04] <wgrant> allenap: too ^^
[10:48] <nigelb> gmb: Great picture of the community team!
[10:48] <gmb> Ta
[10:48] <nigelb> dholbach just told me you did the magic there :)
[11:03] <jml> At the lightning talk, forgot to say "Trust pyflakes"
[11:03] <jml> 'bzr ls -VR --kind=file --null | xargs -0 grep -In %s' is also a good replacement for 'bzr grep'
[12:16] <wgrant> flacoste: sudo -u archvsync /home/archvsync/scripts/schedule-logsync.sh staging
[12:24] <Riddell> StevenK: do you have the group photo?  I'd like to do a blog post
[12:24] <Riddell> or gmb?
[12:25] <bac> Riddell: it is on the wiki, somewhere
[12:26] <bac> Riddell: https://wiki.canonical.com/Launchpad/Sprints/Thunderdome2011
[12:42] <jtv> danilos, spiv: expander landing
[12:42] <jml> hey
[12:43] <jml> there's some scripts out there to automate the Launchpad part of releasing projects
[12:43] <jml> does anyone remember where they are?
[12:43] <jml> poolie: ^
[12:45] <spiv> jtv: yay!
[12:48] <jml> lp:lptools, for those burning with curiosity
[12:53] <jtv> Does reading a feature flag from TAL (tal:condition="features/…") require the page to be on a LaunchpadView?
[12:54] <jtv> I'm getting KeyError: "features"
[12:56] <spiv> request/features perhaps?
[12:57] <jml> any way to automate uploading a tarball & sig for a release?
[12:58] <jelmer> jml: ubuntu-dev-tools has a script for uploading a tarball IIRC
[12:58] <jml> jelmer: ta
[12:58] <jelmer> jml: lp-project-upload
[12:59] <lifeless> also there is the kraken
[13:00] <lifeless> and stuff in lptools
[13:00] <jtv> spiv: no, it's "features/" not "request/features/" — plenty of other templates use it.
[13:00] <jtv> It's not the LaunchpadView though.
[13:00] <jtv> AFAICT this does use one.
[13:00] <lifeless> its mapped as a macro in the base template
[13:01] <lifeless> that expands features to request/features I believe.
[13:01] <jml> lifeless: lptools hides its ability to upload stuff
[13:01] <lifeless> night all, have a good flight back your various homes
[13:01] <jml> I've never heard of the kraken in context of launchpad
[13:01] <jtv> good night lifeless!
[13:01] <jml> lifeless: g'night.
[13:01] <spiv> jtv: grep suggests plenty of other templates use request/features too :)
[13:01] <lifeless> jml: ask abentley about it
[13:01] <jtv> ah :)
[13:08] <benji> huwshimi: https://devpad.canonical.com/~lpqateam/qa_reports/deployment-stable.html
[13:14] <jtv> spiv: https://code.launchpad.net/~jtv/launchpad/bmp-inline-diffs/+merge/66591
[13:15] <jtv> spiv: the feature we're working on would actually have been helpful in producing that branch!
[14:21] <thumper> abentley: are you near sinzui?
[14:21] <abentley> thumper: No, I don't know where he is.  I'm in the downstairs room.
[14:22] <thumper> abentley: ok, thanks
[14:22] <abentley> thumper: Oh, I see him now.
[14:22] <thumper> abentley: could you tell him I want to talk to him about launchpad on oneiric
[14:24] <abentley> thumper: where are you?
[14:24] <thumper> abentley: DX breakout 2nd floor
[14:24] <abentley> thumper: he's on his way.
[14:24] <thumper> abentley: ta
[14:34] <adeuring> deryck: https://code.launchpad.net/~launchpad-orange-squad/launchpad/bug-740208-obfuscate-ws/+merge/66583
[14:36] <StevenK> abentley: https://code.launchpad.net/~stevenk/launchpad/approximateduration-no-words/+merge/66586
[14:38] <abentley> StevenK: A dict indexed by integer makes my head hurt.
[14:39] <StevenK> Yeah .... It isn't very plesant code
[14:51] <wgrant> rvba: ch.queue_declare('bdirect', arguments={"x-expires": 600})
[14:51] <wgrant> So, x-expires: 60000 or so.
[15:05] <bigjools> LONG POLL SQUAD HAS MADE FIRE
[15:07]  * benji wonders if bigjools was slightly confused about his goal.
[15:08] <bigjools> you just need to think of Tom Hanks on the beach of his island getaway
[15:18] <StevenK> allenap: https://code.launchpad.net/~stevenk/launchpad/branch-subscriber-vertical-space/+merge/66615
[15:57] <poolie> woo
[15:58] <jtv> woo?
[15:58] <jtv> woo who?
[15:59] <jtv> …or whom?
[15:59] <nigelb> So, no team photo for launchpad team? :D
[16:03] <poolie> there is one somewhere
[16:03] <poolie> woo FIRE
[16:11] <Riddell> nigelb: see planet.kde
[16:12] <nigelb> Riddell: hah
[16:12] <nigelb> Riddell: tucked away on another planet :P
[16:13] <Riddell> jcsackett: is 421664-code-tab-hover-text forever fated to fail?!
[16:14] <bigjools> allenap: http://paste.ubuntu.com/636395/
[16:14] <nigelb> is bigjools the tall guy on the right?
[16:15] <bigjools> never seen 'im in me life guv
[16:15] <nigelb> heh
[16:16] <nigelb> I can recognize only 6 faces :(
[16:20] <jcsackett> Riddell: I'm going to run those failing tests locally; they don't look like anything your branch introduced.
[16:37] <bac> abentley: my branch is at lp:~bac/launchpad/getnewcache
[16:39] <abentley> bac: cool, thanks!
[16:41] <jcsackett> Riddell: wasn't able to directly land your branch, so its out through the normal process again.
[16:41]  * jcsackett crosses fingers.
[16:49] <StevenK> spiv: Can haz bug about old FF removal?
[16:49] <StevenK> spiv: It should be fairly easy to rip it out -- and I'll be bored on the plane ...
[17:26] <spiv> poolie: http://people.canonical.com/~andrew/Inline-diff-screenshot.png
[17:26] <poolie> ta
[22:14] <wallyworld> wgrant: meet at 7:30 tomorrow for breakfast?
[22:40] <wgrant> wallyworld: Sure!
[22:42]  * wgrant sleeps.