[00:05] thumper (and sinzui), but 664220 === Ursinha is now known as Ursinha-afk [00:05] bug 664220 [00:05] <_mup_> Bug #664220: Person search is broken on edge === Ursinha-afk is now known as Ursinha [00:26] mars: bug 664220 [00:26] <_mup_> Bug #664220: Person search is broken on edge [00:26] bah [00:26] mars: bug 664220 [00:26] <_mup_> Bug #664220: Person search is broken on edge [00:26] bah. [00:26] I've duped it anyhow [00:28] lifeless, thanks. dupe search did not turn up that bug [00:28] neither did google [00:28] mars: bugs.lp.net/launchpad-project/+bugs?field.tag=timeout [00:28] I always hit that up and eyeball these days === Ursinha is now known as Ursinha-afk [00:54] Project devel build (138): STILL FAILING in 3 hr 31 min: https://hudson.wedontsleep.org/job/devel/138/ [00:54] * Launchpad Patch Queue Manager: [r=gmb][ui=none][bug=347218] Allow bug supervisor to set [00:54] official_bug_tags for a distribution. [00:54] * Launchpad Patch Queue Manager: [r=lifeless][ui=none][no-qa] Fix multiline glob imports in doctests. [00:54] * Launchpad Patch Queue Manager: [r=mars][ui=none][bug=648476] Adds the "Add member" link to the block [00:54] that displays in the member listing when there are no members; [00:54] team owners should still have access to this link, regardless of [00:54] membership status. [00:54] * Launchpad Patch Queue Manager: [r=abentley][ui=none][no-qa] Fix bin/kill-test-services to work [00:54] properly again [00:54] * Launchpad Patch Queue Manager: [r=sinzui][ui=none][bug=655802] Fixed timeout when AJAX picker [00:54] searched person or team vocabularies. [00:54] * Launchpad Patch Queue Manager: [r=gmb][ui=sinzui, [00:54] salgado][bug=652232] Move links into sidebar on code.lp.net/$person [00:54] page. [01:02] rockstar: ping - quick question [01:02] wallyworld, whazzup? [01:03] Oh balls. Just realized it's time for the AsiaPac reviewer's meeting. [01:03] the registrant attribute on IBranchMergeQueue..... [01:03] says "The user that registered the branch" [01:03] that isn't correct is it? [01:04] wallyworld, no, it's not. I apparently can't always finish my thoughts. [01:04] looks like a cut'n'paste :-) [01:05] btw, i thought the reviewers' meeting time had changed? [01:10] bdmurray: https://devpad.canonical.com/~lpqateam/qa_reports/deployment-stable.html [01:10] lifeless: so, it really is okay just incomplete rather [01:11] lifeless: it only allows it for a project and I've another branch that allows it for a distribution [01:13] bdmurray: please mark it as qa-ok then [01:13] cause its blocking deployments [01:15] we need to iterate on our toolchain to allow more precision [01:23] Project db-devel build (88): STILL FAILING in 3 hr 35 min: https://hudson.wedontsleep.org/job/db-devel/88/ [01:23] Launchpad Patch Queue Manager: [rs=buildbot-poller] automatic merge from stable. Revisions: 11748, [01:23] 11749, 11750, 11751, 11752, 11753, 11754, 11755, 11756, 11757, [01:23] 11758, 11759, 11760, 11761 included. [01:23] oh hai. === Ursinha-afk is now known as Ursinha [01:30] Dammit PQM... [01:47] rockstar: whazzup? [01:47] rockstar: thats more buildbot :) [01:48] thumper, All lines of log output:"[['Connection closed: Unexpected end of message. Please check connectivity and permissions, and report a bug if problems persist. ']]" [01:48] :( [01:48] lifeless, ^^ That error is not buildbot... [01:49] rockstar: the irc message is hudson [01:49] the size of the change is buildbot [01:49] lifeless, yeah, I can /ignore LPCIBot. :) [01:49] the failure root cause is probably the subunit bug I linked here this morning [01:49] that it doesn't support filtering the bogus errors from zope.testing [01:51] So this is great. I submitted two pipes of a pipeline to pqm. pqm pukes on the first one, but seems to gobble up the second one. All the changes are landed, but now the bug the first pipe was linked to isn't going to find the QA-bot. [01:52] Awww [01:52] rockstar: This is why I feed branches at PQM piece-meal [01:53] rockstar: by pukes [01:53] rockstar: do you mean 'we were in testfix' ? [01:53] rockstar: if you used --fixes lp:xxx, then the qabot should hopefully find it anyway. [01:53] lifeless, no, see my message to thumper above. [01:53] oh, I see [01:53] so codehosting was down [01:54] StevenK, I did feed them piece-meal, in that I sent them both separately. [01:54] spm: would like to know if we have a log of what went on for rockstar above [01:54] lifeless, yeah, it's not necessarily pqm's fault, but this camel is really tired of straw... [01:54] hm? [01:54] rockstar: Sorry, what I mean is submit ; wait for the e-mail until it says sucess and then submit the next one [01:54] rockstar: would tarmac handle that any better? [01:55] lifeless, not really "better" per se. It won't merge the branch if the dependent branch hasn't been merged yet. [01:55] (not that its pqm vs tarmac, this is all about operational polish) [01:55] rockstar: thats intersting for me, because I often want to land all the branches at once. [01:56] rockstar: is there a knob for controlling htat [01:56] lifeless, no, there isn't. [01:56] spm: 13:48 < rockstar> thumper, All lines of log output:"[['Connection closed: Unexpected end of message. Please check connectivity and permissions, and report a bug if problems persist. ']]" [01:56] spm: usually means ssh handshaking didn't. [01:56] mmm [01:56] indeed [01:56] lifeless, I'm sure we can, but I think we can probably teach launchpad to have a better knob for controlling it. [01:57] * rockstar quits for the night instead of getting his grump on [01:58] gnight [02:02] hmm, need the fire on I think. [02:02] brb [02:15] Shouldn't a UI designer be doing UI reviews? [02:18] There was a proposal that we would be able to forward things to the design team. [02:18] But that never happened. [02:18] And it doesn't help overall UI consistency. [02:18] Which Launchpad is sort of completely lacking. [02:26] its not completely lacking [02:26] I think thats a bit harsh [02:27] True, it is better than it used to be. [02:48] thumper: you around? [02:48] yep [02:49] wanna have that skype call? [02:51] thumper: ^^^^^ - i also had a question about that +branch bug [02:51] wallyworld: sure [02:52] wallyworld: skype doesn't like me [02:52] thumper: so it's not the only one then? [02:54] * lifeless hates on doctests some more [02:54] I can hear you [02:54] thumper: can you hear me? [02:55] ok [02:55] this is weird [02:55] ah no, its not. Its subunit [02:55] jml: I think we're going to have to change the zope behaviour here. [02:55] jml: unless you want to filter in the ec2land code [02:59] mwhudson: hey [02:59] lifeless: hello [02:59] my uniqueconfig branch bounced [03:00] I have a fix but its arguable [03:00] gar, lp-service just did too :( [03:00] I want to see if you'd care [03:00] ffs [03:00] lifeless: link me up [03:00] lp:~lifeless/launchpad/uniqueconfig 11748 [03:00] mwhudson: this isn't the thread thing [03:00] ah ok [03:00] I'm going to have a poke at that now [03:01] * mwhudson looks [03:02] the error was testrunner_12345 [03:02] instead of testrunner [03:04] lifeless: it's not very nice, but i think it's good enoguh [03:04] thanks [03:04] Spelling it out long ways was incomprehensible [03:05] not to mention redundant [03:16] mwhudson: ok, threading thing upcoming [03:19] mwhudson: https://code.edge.launchpad.net/~lifeless/launchpad/threads/+merge/39009 [03:23] lifeless: oh, before i do the second clicky, did you make a branch of zope.testing for this based on lp:~mars/zope.testing/3.9.4-p1 ? [03:23] no [03:23] we don't want this upstream [03:23] so that would be pointless [03:23] well, it might make life ever so slightly easier if someone wants to make a 3.9.4-p3 version for whatever reason [03:24] tar xzf ... [03:24] (no really) [03:24] I'll probably get tag filtering done on the flight [03:24] and we can back this out then [03:25] that makes it a bit harder for the next hacker to use version control [03:25] i've done stuff like that and then been freaked out when bzr diff didn't work :) [03:26] the thing is that this is already out of vc [03:26] if we were using branchs ala config-manager [03:26] I'd totally agree [03:26] if you like I can do it [03:26] it just seems odd [03:26] particularly for a short term (hopefully) hack [03:27] i would rather you did, but don't want to make too big a fuss [03:27] compromise [03:27] and you know the thing about short term hacks... (although i agree in this case it probably will be) [03:27] if after uds I haven't done the real deal [03:27] I'll do it [03:27] k [03:28] mwhudson: lucilleconfig was added as a temporary hack until the schema was sorted out. [03:28] 6 years and a few days ago. [03:28] It is still there. [03:28] wgrant: NOT HELPING [03:28] lifeless: approved [03:28] NO TEMPORARY HACKS. [03:28] mwhudson: thanks [03:28] mwhudson: now, should I ec2land it :P [03:28] eh [03:28] probably i guess [03:28] yeah, I was trolling === Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha === Ursinha is now known as Ursinha-afk [04:11] mwhudson: it seems windmill still doesn't like me [04:11] jam: yeah, i gave up and just lp-landed it [04:11] r [04:24] mwhudson: thank you [04:25] I ♥ addDetail [04:25] lifeless: you ♥ unicode, too, I see [04:27] lifeless: damn, how do I type that? [04:27] * jtv gotta have [04:32] Project devel build (139): STILL FAILING in 3 hr 38 min: https://hudson.wedontsleep.org/job/devel/139/ [04:32] * Launchpad Patch Queue Manager: [r=edwin-grubbs][ui=salgado, [04:32] sinzui][bug=652156] Updates the projectgroup branches page to show a [04:32] message consistent with the other codehosting_usage messages [04:32] when there are no branches, instead of an empty table. [04:32] * Launchpad Patch Queue Manager: [r=jtv][ui=none][bug=659050, [04:32] 663075] Only load the link bug and subscription ++form++ when the [04:32] user is logged in. [04:32] * Launchpad Patch Queue Manager: [r=mwhudson][ui=none][bug=624009] Only send the initial review email [04:32] when the merge proposal needs review. [05:24] statik: oh hai :) [05:35] \o/ [05:35] \o/ [05:35] \o/ [05:35] \o/ [05:35] \o/ [05:35] \o/ [05:35] A [05:36] Oh? [05:36] That bad? [05:37] that good [05:37] totally ephemeral librarian [05:37] Nice! [05:37] It even finds its own ports to lurk on? [05:38] yes [05:38] and root dir [05:39] it also shows pretty clearly that we have a leak on the librarian :) [05:39] not in, of [05:39] lp:~lifeless/launchpad/librarian if you're interested [05:52] thumper: trade you reviews? [06:09] Project db-devel build (89): STILL FAILING in 3 hr 58 min: https://hudson.wedontsleep.org/job/db-devel/89/ [06:11] lifeless: I'm done for today... === almaisan-away is now known as al-maisan [08:02] Project devel build (140): STILL FAILING in 3 hr 29 min: https://hudson.wedontsleep.org/job/devel/140/ [08:02] * Launchpad Patch Queue Manager: [r=mwhudson][ui=none][bug=660264] Implement LaunchpadForkingService to speed up bzr+ssh connection times. [08:02] * Launchpad Patch Queue Manager: [r=edwin-grubbs][ui=none][bug=663828] Split ReadyService into its own file for easier maintenance of TacTestSetup. [08:46] good moring [09:15] Hello [09:32] lifeless: the ec2 land code is complex enough already... and I've seen that you've changed testrunner, so it's all good [09:41] bigjools: How long has p-dr been broken? [09:42] It should have died within a day of the Intrepid purge, if that was it. [09:42] And I can't think what else it could have been. [09:42] :/ [09:43] Project db-devel build (90): STILL FAILING in 3 hr 33 min: https://hudson.wedontsleep.org/job/db-devel/90/ [09:43] Launchpad Patch Queue Manager: [rs=buildbot-poller] automatic merge from stable. Revisions: 11762, [09:43] 11763, 11764 included. [09:56] wgrant: I don't know. A Long Time. [10:01] Ah, good. === matthew_ is now known as mrevell [10:49] wgrant: can you remember if there was another bug that left buildds idle? [10:49] not the aborted/aborting one that I fixed [10:49] artigas/hooker are not happy [10:49] bigjools: What's their status? [10:49] idle [10:50] The slave too? [10:50] lots of these in their logs: [10:50] 2010/10/21 10:19 +0100 [HTTPChannel,202,91.189.90.177] 91.189.90.177 - - [21/Oct/2010:09:19:53 +0000] "POST /rpc HTTP/1.0" 200 222 "-" "xmlrpclib.py/1.0.1 (by www.pythonware.com)" [10:50] artigas is sparc, hooker ia64, right? [10:50] yarp [10:50] bigjools: that just means "Successful XMLRPC request" [10:50] quite [10:50] bigjools: Calling the status() method says BuilderStatus.IDLE? [10:51] who knows, that's not something that's easy to find out [10:51] bigjools: fwiw, it wouldn't be too hard to get the actual methods called in the log. [10:51] we restarted the buildds and they're still idle [10:51] bigjools: There are no errors in the log about the directory already existing? [10:52] And it's not the b-m thing where it ignores new builders until it's restarted? [10:52] nope [10:53] Well, I'd be hijacking cesium for a quick python -c "import xmlrpclib; xmlrpclib.ServerProxy('http://artigas.buildd:8221/rpc').status()" [10:56] yeah I might do that [10:59] wgrant: it's IDLE [11:00] as expected [11:00] bigjools: What does b-m say about it? [11:00] zip [11:00] I'm suspecting the usual b-m bug. [11:00] Except that these builders aren't new. [11:00] I'm not sure that is a bug [11:01] win 67 [11:01] lifeless: £ or $ ? [11:01] bigjools: Isn't it? [11:02] wgrant: I had another instance yesterday of the b-m throwing lots of connection errors, but no builder had been added, so my original assessment was probably wrong [11:02] this is something different again [11:03] Yay [11:03] Maybe it will mysteriously vanish with builderslave-resume. [11:04] best branch name evar [11:05] * bigjools considers renaming it to buildd-manager-apocalypse [11:05] jml: how can we get the methods in the log, BTW? [11:06] bigjools: off the top of my head, I don't know. it's a matter of finding out what does the dispatch to xmlrpc_FOO and whacking a log in there. [11:06] or just putting log statements in the actual xmlrpc_FOO methods [11:07] yeah that was what I would do, I thought you had a "clever" way :) [11:16] one of these days I'll write a decorator that logs function calls and return values [11:22] I don't understand why people like the GitHub "Fork" thing... it makes no sense :/ [11:23] it makes a little bit of sense [11:23] I wish Launchpad had server-side branching [11:23] Why? [11:23] Because it's a shedload quicker for Launchpad to branch Launchpad than it is for you too [11:24] StevenK: well, that's only true if you don't ever intend to work on your branch [11:25] although "branch to local" then "push to LP" is probably slower than "branch on server", "branch to local" [11:25] Modulo bzr bugs it shouldn't be significantly slower. [11:25] wgrant: sometimes it's more a way of thinking than anything else, I guess. [11:25] (I know that in practice it is) [11:27] you know, I'm not grateful enough for the fact that I'm not an IP lawyer [11:27] jml: The Apple thing? [11:28] wgrant: well, that's the trigger, but it's also every damn thing about IP law [11:28] Heh, yes. [11:28] It is messy stuff. [11:28] I just want to make stuff, pay artists for stuff I like and get to use the stuff I buy. [11:29] And not have Apple steal your product names? :) [11:29] jml: any word from legal? [11:30] lifeless: no, none yet. [11:32] Project devel build (141): STILL FAILING in 3 hr 30 min: https://hudson.wedontsleep.org/job/devel/141/ [11:32] Launchpad Patch Queue Manager: [r=abentley][ui=none][bug=114766] Restrict the ability to nominate [11:32] bugs for a release to bug supervisors, owners or drivers. [11:33] :( [11:33] * StevenK notes that hudson has gotten quicker [11:35] StevenK: oh is the time in the IRC message the time for the build? [11:37] It took me a couple of days to work that out. [11:38] jml: you have mail. I'd appreciate you taking the time to read it thoughtfully ;) [11:38] jml: you'll understand why I specifically mention this when you see its size! [11:38] lifeless: ok :) [11:40] jml: I think we should filter in ec2test, but I'll do prep to make it easy, in subunit. [11:41] lifeless: well, lib/devscripts/ec2test/remote.py is the relevant place. [11:42] lifeless: also, importantly, lib/devscripts/ec2test/tests/test_remote.py [11:53] I need an incremental eyeball [11:53] rev 11773 of lp:~lifeless/launchpad/threads [11:54] lifeless: ok. [11:55] this should, finally, fix the test environment headaches [11:55] with threads [11:55] anything to distract me from email [11:55] + new_threads = new_live_threads() [11:56] lifeless: that's the only interesting change in that revision [11:56] thats it [11:56] if you look at the commit message, or the context, it should make more sense [11:56] lifeless: well, that's a perfectly syntactically formed line of Python :) [11:56] thank you [11:56] lifeless: fetching the branch now to get context [11:57] I spent hours on it [11:57] jml: loggerhead can show that ;) [11:57] lifeless: loggerhead is too slow. [11:57] hah, ouch, [11:57] Loggerhead still times out for me on the first request for any LP branch. [11:57] But apart from that it's not too slow. [11:59] wgrant: it's not significantly faster than me fetching the branch and using my editor. [11:59] wgrant: especially given the way it always times out on the first request for any LP branch [11:59] lifeless: +1 [11:59] Morning, all. [12:00] jml: Ah, I guess it's different since you live next to the DC. [12:00] jml: Yes, 'in 3 hours 3 minutes' is how long it took [12:01] jml: I've thrown it straight at PQM [12:01] StevenK: I read it as "will still be failing in 3 hours 3 minutes" [12:01] jml: Patches welcome [12:01] jml: if it doesn't get through you probably want to resubmit it yourself: '[r=mwhudson][ui=none][bug=663644][no-qa] Switch to a patched zope.testing that does not emit subunit errors on leaked threads, for windmill test sanity.' [12:02] \o/ \o/ [12:02] (I think we may be in testfix) [12:02] lifeless: Was the patch to zope.testing horrible? [12:02] no its tiny [12:02] we'll get a longer term fix later [12:03] hmm, shoving testfix on that [12:03] lifeless: would you like to borrow some of my apostrophes? [12:03] because it does, thought a different one. [12:03] jml: prhaps [12:05] jml: you're turning into mpt [12:05] (not a bad thing) [12:05] lifeless: my penmanship is vastly inferior. [12:06] but your list of ways to improve things are equivalent [12:06] given enough pens and enough jmls, the penmanship could be magnificent [12:06] or the universe may end [12:07] think of all the hair [12:07] StevenK: whats the thread topic you started about threading errors in hudson ? [12:07] * bigjools goes back to fixing p-d-r [12:07] found it I think [12:08] lifeless: Failures on Hudson [12:08] bigjools: Which files are screwed? [12:08] querying that as you type [12:08] Ah, so that kind of fixing. [12:09] I'm not sure what could have caused this [12:09] The Intrepid purge could easily have, if the timing was just right... [12:09] other than over-zealous librarian file collection [12:09] But that seemed unlikely. [12:10] we probably marked something expired when it should not have been [12:10] * bigjools would love reference counts on LFAs [12:10] 01:17 So, this query isn't perfect. It violates the PPA blacklists and stay of execution. But that should all be long-dead for intrepid. [12:10] hmm [12:15] wgrant: they are all gutsy files [12:16] bigjools: What's datesuperseded/dateremoved? [12:17] Sources or binaries? [12:17] only checkiungbinaries [12:17] sigh [12:17] only checking binaries [12:17] Er, yes, of course. [12:18] all 90 are superseded at the same time: 2010-05-28 14:12:50.17775 [12:20] Not Deleted? [12:20] so if I update these to all have dateremoved set then it should fix p-d-r [12:20] bigjools: LFAs are not really intended to be shared with other objects [12:20] bigjools: does soyuz share them ? [12:20] But it would also leave cruft all over the disk. [12:20] lifeless: it shares them across publications [12:20] (LFAs are the reference counts for content objects) [12:21] gnight [12:21] wgrant: point [12:21] nn lifeless [12:22] it's only 90, I can collect filenames and do it manually [12:22] bigjools: What's the expiry date on the LFA? [12:23] 2010-01-07 [12:23] ! [12:23] ... pardon? [12:23] * bigjools scratches head [12:23] Is datepublished set? [12:24] yes [12:24] :( [12:24] I suspect we set expires back a long way to force GC action [12:25] Hmm. [12:25] that date is close to our sprint in NZ [12:25] It is, yes. [12:25] And I found a bug in that query soon after. === al-maisan is now known as almaisan-away [12:26] So it's fine. Delete them manually. [12:27] (the January query included in the EXCEPT clause only those binaries outside (feisty, gutsy), ignoring the possibility that (feisty, gutsy) pubs could still be alive outside the primary archive) [12:28] yeah [12:28] people backporting no doubt [12:28] I wonder if we could do what lifeless suggests [12:28] would make life easier [12:28] Not easily. [12:28] Without redesigning most of the data model in unobvious ways. [12:29] (that date does match: the paste of the query is from the 2010/01/14, and it backdated expiry by a week) [12:29] well if it gives us gains elsewhere it's worth it [12:29] By "unobvious" I mean that it's not obviously possibly at all. [12:30] it's always possible [12:30] Not without making everything even uglier. [12:30] so pessimistic already at such a tender age ... [12:30] Shh. [12:30] :) [12:31] it took me years to get this jaded [12:37]