[01:21] hmm, I'm seeing weird behaviour talking to the qastaging server [01:22] retrieving a wadl resource I sometimes (I don't see a pattern yet) only get 3486 bytes back [01:22] How big are the headers? [01:22] And what is the content? [01:23] the content is JSON (but cut off halfway through, so the JSON parser barfs) [01:23] We occasionally see that when production is overloaded. [01:23] The response, including headers, gets truncated to a power of two. [01:23] I do not remember which power of two. [01:24] it seems to consistently be 3486 here, and it appears to happen randomly [01:24] I wouldn't be surprised if the size of the headers was (4096-3486) [01:24] That sounds about right. [01:24] Which resource? [01:24] Authenticated or not? [01:25] yes, should be authenticated [01:25] would anonymous access make a difference in any way? [01:26] Probably not on qastaging. [01:30] do you know if there is a bug of some sort? [01:30] ehm, bug /report/ [01:32] jelmer: Bug #636713 [01:32] <_mup_> Bug #636713: during sustained overload situation replies are truncated in production < https://launchpad.net/bugs/636713 > [01:33] wgrant: thanks! [01:33] unfortunately my script can only run against qastaging at the moment, so I'll try again later [01:35] Why can it only run against qas? [01:35] it uses some new stuff that Ian introduced recently [01:35] http://samba.org/~jelmer/recipe-status/ [01:36] It's not even on staging? [01:36] has some of the output [01:36] ah, the new recipe build collections? [01:36] yep [01:36] it wasn't on staging a couple of days ago, let me check again.. [01:36] Ah, staging's update failed overnight. [01:36] Will look tomorrow. [01:36] Too many incidents already this weekend. [01:37] thanks, it does actually appear to be on staging now [01:39] Odd. [01:39] I was expecting staging to be down. [01:40] Hmm. [01:41] jelmer: You may be interested to know that they are on prod now. [01:42] But they only show up in 1.0, not devel... [01:43] Ah, no, they're in devel too, but you have to kill your cache. [01:43] oh, cool [01:43] they don't show up in launchpad.net/+apidoc/devel [01:45] They do if you avoid your cache. [01:45] eg. Ctrl+Shift+R. [01:46] oh, wow. I didn't realize Launchpad did heavy caching on that too. I thought it was just the wadl. [01:46] Neither did I. [01:46] But I thought it was worth a try. [01:48] jelmer: isn't https://bugs.launchpad.net/bugs/725811 a dupe? [01:48] <_mup_> Bug #725811: expose ISourcePackageRecipeBuild.binary_builds on the web UI < https://launchpad.net/bugs/725811 > [01:49] lifeless: a dupe of which other bug? [01:49] ians work [01:49] No. [01:49] lifeless: No, Ian exposed the source package builds [01:49] can you not traverse to the binaries ? [01:49] this is about the binary builds created from those source builds [01:50] yes, I get that [01:51] lifeless: I can't find any way to traverse to the binaries, the source_package_recipe_build doesn't have anything that links it to the source package it created [01:52] We should probably guess at that. [01:52] wgrant: the source_package_recipe_build has a binary_builds attribute that's not exposed [01:53] I currently parse the build log for the source package name and the version, and then call archive.getPublishedSources(source_name=source_name,version=version)[0].getBuilds() [01:53] jelmer: Ah, so it already does guess. [01:53] We should expose that and the guessed SPPH. [01:53] wgrant: yep - the web page already shows that information [01:54] Yay, fpc built correctly on i386 and amd64 this time. [01:54] at least it works against production now :) e.g. http://samba.org/~jelmer/recipe-status/testing-cabal.html [01:54] wgrant: fpc? [01:54] jelmer: The last bit of fallout from https://wiki.ubuntu.com/IncidentReports/2011-02-25-Permissions-build-failures [02:00] woo - statement count gropuing useful again [02:00] Oh? [02:00] you've killed enough of the checkwatches stuff [02:00] 1992 OOPS-1883CCW1185 https://bugzilla.novell.com/show_bug.cgi?id=604251 [02:00] Almost. [02:00] is still in the list [02:00] Still got lots of ProtocolErrors. [02:00] Hmm. [02:00] That's odd. [02:00] Looking. [02:00] It should be mostly bugtracker-wide errors now. [03:51] Woah. [03:51] Looks like LP now uses the Light variant of the font... [04:41] hmm [04:41] I -really- have to fix this design issue preventing test optimisation with baseline fixtures. [04:42] making a new keyspace and schema with cassandra is a little slow to do per-test. === timrc- is now known as timrc [09:00] Project db-devel build #401: STILL FAILING in 5 hr 48 min: https://hudson.wedontsleep.org/job/db-devel/401/ [14:51] The front-page of staging OOPSes - yay :-( [17:57] maxb: what is hte oops ids [18:04] OOPS-1884S112844 for example, but it's entirely repeatable [18:05] geser mentioned an "S" OOPS id in #launchpad recently too [18:07] maxb: the reason I asked for anexisting one is that the oops only sync every $hours on staging [18:07] looking at gesers [18:07] https://api.staging.launchpad.net/1.0 [18:14] lifeless: o/ [18:14] oh hai [18:14] lifeless: so I wanted to setup a bug tracker for the packages... [18:15] SpamapS: upload them once [18:15] and at one time I did have the project cassandra-packages registered [18:15] but now its like that project disappeared [18:15] I can't create a new cassandra-packages ... but I can add branches of it. [18:15] 2010-06-16 Curtis Hovey: Project disabled and Clint Byrum (Clint Byrum) notified [18:16] I can reenable it [18:16] Doh maybe I missed that notification [18:16] but why not just upload to natty [18:16] its not suitable [18:16] that will activate the package bug tracker [18:16] thrift is embedded.. [18:16] avro.. [18:16] ok [18:16] :( [18:16] we're a lot closer to distro packages... but there are still things embedded. :-P [18:17] * SpamapS notes that the .debian.tar.gz is bigger than the .orig.tar.gz [18:18] reenabled [18:18] I suggest a more clear description and title [18:18] 'e.g. 'staging for cassandra packages' and 'cassandra embeds too much stuff to be blah blah blah [18:19] you'll want to configure the bug tracker [18:19] SpamapS: ^ [18:21] SpamapS: it was a bit disapointing that cassandra can't do schema changes in *separate keyspaces* concurrently :> [18:21] SpamapS: I have some /awfully/ crufty code you'll want to see :> [18:23] lifeless: everybody using cassandra has some awfully crufty code to deal with their still-evoling admin capabilities [18:23] OTOH I have most of a cassandra based oops system [18:23] but their inability to do range requests is plain annoying [18:23] I've mailed our contact at riptano again :) [18:24] wait ranges for reads are supported [18:24] as long as you use OPP [18:24] which, given the OOPS focus, would make sense, and be only slightly confusing. [18:24] not on secondary indices because they are bound to the node of the primary row [18:25] right. not much you can do there. [18:25] sure there is; they can fix the broken assumption :> [18:26] for now I've use a trick [18:26] you have a column with value '' [18:26] and you index it [18:26] then you also index the thing you want ranged [18:26] and you do a search for '' and $range [18:27] it reads every row and does the range stuff as a loop [18:27] this will suffice for a while [18:27] Right.. is that worth it? [18:27] given I only need that for gc [18:28] for aggregates and reports I'm keeping running totales [18:29] there's some rudimentary count suppot in 0.7 .. and 0.8 is supposed to have powerful counting [18:29] one of the weirdnesses [18:30] in twissandra they use a timestamp to build the timeline [18:30] but timestamps - even @ microseconds - are not unique [18:30] it might be /hard/ to collide, but get enough activity and frontends and it will happen [18:31] yeah there's some work on making it a vector clock instead of a straight microsecond timestamp [18:31] see, I could ues that [18:31] the other thing I could use is a ttl [18:31] but open bug references need to impede gc [18:31] I'm thinking of having two cfs [18:31] identical [18:32] but one referenced by bugs and no ttl [18:32] that would leave aggregated cruft behind though, because AIUI I wouldn't get a callback on ttl expiry [18:33] Couldn't you just remove the ttl when a reference is reated? [18:33] wasn't 100% sure that doing a new write with no ttl would do that [18:33] and was coding at us unfriendly times [18:33] SpamapS: so I reactivated https://launchpad.net/cassandra-packages [18:34] SpamapS: but you need to click on the bug tracker and configure it [18:34] it thinks you don't use LP at the moment [18:34] lifeless: thanks. :) I looked in my nbox and I must have inadvertently deleted the notification [18:38] are you familiar with the bug settings bit I'm refering to ? [18:38] very [18:38] kk [18:38] just having dinner now.. somebody reminded me that there is no food to be had in capetown this late on sunday ;) [18:39] I'm just polling till I can file these bugs for you :) [18:39] EAGAIN [18:43] lifeless: ok, bug reporter configured [18:50] cool [18:51] want all the things I mailed you as bugs? [18:54] SpamapS: ^ [18:54] yes please! [18:55] I think some of them will be solved by doing actual lucid/maverick backports rather than copying the bin packages [19:00] done [19:00] SpamapS: ^ [19:00] awesome... I sent my reply as well [19:00] to your email I mean [19:00] cool [19:01] These Afrikaaners have filled me up with delicious food.. hopefully I'll pass out soon. :) [19:03] nice [19:06] its good to see that you're making some headway w/ the cassandra packages. [19:06] I was beginning to wonder if anybody would ever actually use them. :) [19:08] heh [19:08] so the only critical one is the classdefnotfound [19:08] not being able to compact is, uhm, operationally significant [19:12] indeed [19:12] $ ./get-ppa-stats cassandra-ubuntu stable [19:12] cassandra 0.7.0-0ubuntu1 42 [19:12] 42 downloads.. not too shabby ;) [19:25] Where does get-ppa-stats live? [19:28] in my home dir ;) [19:29] actually [19:29] lp:~clint-fewbar/+junk/lptools [19:29] maxb: ^^ [19:30] lifeless: one point of pain right now is that they up and changed the build system entirely from 0.7.0 -> 0.7.1 [19:37] thanks [19:38] SpamapS: of course they did [19:39] SpamapS: get-ppa-stats should be in lptools ;) [19:43] ok.. passing out eminent [19:50] night [20:16] hi flacoste [20:17] hi lifeless [20:19] flacoste: I spiked a oops-in-cassandra in the weekend; its on my local disk only at the moment [20:19] flacoste: ok with you if I make a project in lp, agpl3 the code and push publicly? [20:19] flacoste: also, how was your flight? [20:20] lifeless: fine by me [20:20] lifeless: my flight was awesome, even though it started by a cancelling flight [20:20] >< [20:20] cancelled [20:20] so the replacement they upgraded you or something ? [20:21] lifeless: kind of, instead of flying continental to London, i got an air canada flight [20:21] oh nice [20:21] and i managed to use one of my upgrade ticket [20:21] \o/ [20:21] so i first business for the first time [20:21] slept better than at home, 5 hours straight without interruptions [20:21] no kids :) [20:21] exactly [20:21] :-) [20:22] and on the london cape town flight, it wasn't fully booked, so i got an empty seat besides me [20:22] which made sleeping also easier [20:22] so one of the best flight experience overall! [20:22] feel fresh and ready for this week sprint [20:24] excellent [20:24] we have fresh water again, came on on friday [20:24] :) [20:25] that's always useful [20:25] very [20:33] all software sucks ;) [21:03] thumper: mumble? [21:07] wallyworld: two bugs you might like to know about - model defects exposed by your nice patch to show merge proposals in the revision lists for branches [21:07] https://bugs.launchpad.net/launchpad/+bug/726190 [21:07] <_mup_> Bug #726190: bugs shown against merge proposals in revision list lists all bugs ever linked to branch and differ to those the MP itself show < https://launchpad.net/bugs/726190 > [21:07] https://bugs.launchpad.net/launchpad/+bug/726195 [21:07] <_mup_> Bug #726195: merge proposals show closed bugs < https://launchpad.net/bugs/726195 > [21:13] rhythmbox just rickrolled me [21:27] mornign wgrant [21:27] hmm, no sinzui [21:29] anyone know if there is a bug about things like ubiquity showing up in +needs-packaging at all ? [21:30] lifeless: What's special about ubiquity? [21:30] It still has a project, doesn't it? [21:31] no [21:31] you may be htinking unity [21:31] dpkg, ubiquity, apport - those things the distro folk often explicitly do not want 'upstream' projects [21:31] Erm. [21:31] Have you seen devel lately? [21:32] It needs a new apt-ftparchive that is not everywhere yet. [21:32] I suspect we want to roll that back. [21:32] you just changed topic, right ? [21:32] I did. [21:33] So, there is a bug about how we don't really support Ubuntu-native packages well at all. [21:33] for context [21:33] I'm looking at https://bugs.launchpad.net/launchpad/+bug/722794 [21:33] <_mup_> Bug #722794: DistroSeries:+needs-packaging timeouts < https://launchpad.net/bugs/722794 > [21:33] which does crazy stuff to get bugs and strings [21:33] and bug heat?! === Peng__ is now known as Peng [22:10] lifeless: thanks for the heads up on those bugs - that post implementation clean up stuff is on my radar for this maintenance phase [22:11] wallyworld: the timeout I mentioned last week preexisted [22:11] wallyworld: I have a branch playing to amliorate it [22:11] lifeless: yeah, i had a quick look and had come to the same conclusion [22:11] wallyworld: https://code.launchpad.net/~lifeless/launchpad/bug-710685/+merge/51258 [22:12] was going to get in today and try and fix it but you already are on it [22:16] Uh what? 0 tests run in 4:12:24.883708, 0 failures, 0 errors. It took four hours to do exactly what then? [22:16] Haha. That's the question. [22:21] huwshimi: Bug #726243 [22:21] <_mup_> Bug #726243: ec2 no longer keeps track of failed tests < https://launchpad.net/bugs/726243 > [22:24] wgrant: Oh, that's not very helpful. [22:25] It does at least detect whether it has failed or not, though. [22:27] wgrant: How do I know what failed then? The log is not obviously showing anything that went wrong, but there's a lot of the log that I haven't looked at :) [22:27] huwshimi: You'll need to grep it. [22:27] Or run subunit/testr over it [22:27] wgrant: For what? [22:30] I presume there's some subunit tool, but I tend to search for "error:" and "failure:", or just pipe it into 'testr load'. [22:32] Grab the log from the e-mail and: zcat | subunit-filter --no-passthrough | subunit-ls [22:33] I'm guess it is this: http://paste.ubuntu.com/573231/ [22:33] wgrant: subunit *tool*? Why have one confusing UI when you can have ten! [22:34] huwshimi: I fixed that on Friday. [22:34] huwshimi: If it's the only failure, lp-land it. [22:34] Except that we're in testfix. [22:34] Because buildbot doesn't have the new apt. [22:34] And I guess we are ISless today. [22:34] wgrant: Ah right, it appears to be. Do I need to merge devel first? [22:35] No. [22:35] wgrant: the new apt is in our ppa [22:35] wgrant: IS will be here when spm shows up [22:35] Given he flew back from London on Saturday? You tell funny jokes. [22:35] lifeless: Right, but given that he may well still be in the air... [22:36] he should get in at 9am canberra time or thereabouts [22:36] and he will want to stay up to beat jetlag [22:36] [whether he will succeed... different question] [22:37] Heh [22:39] lifeless: What does subunit.TestProtocolServer do? You give it a result object and a stream, and it plays the stream's events through the result? [22:40] Ah, no, that stream is something else... [22:42] wgrant: pydoc subunit [22:43] ' Subunit has support for non-blocking usage too, for use with asyncore or [22:43] Twisted. See the ``TestProtocolServer`` parser class for more details. [22:43] ' [22:43] wgrant: what do you want to achieve? [22:43] lifeless: Just working out how ec2test.remote does its stuff. [22:44] So I can unbreak it. [22:44] ok [22:44] well - class TestProtocolServer( - find that section in pydoc [22:44] I just reread it and it seems useful to me :) [22:46] BranchSet:CollectionResource:#branches never completes in < 4 seconds :( [22:53] Project devel build #481: STILL FAILING in 5 hr 56 min: https://hudson.wedontsleep.org/job/devel/481/ [22:53] StevenK: Jenkins needs the new apt. [22:54] StevenK: And you need to tell me what the checkwatches.txt failure is, since it 404s in the web UI... [22:57] Yes, it doesn't like doctests. [22:58] :( [22:59] http://pastebin.ubuntu.com/573238/ [22:59] Thanks. [22:59] Just checking the current build slave out [23:00] ubuntu@ip-172-56-124-250:~$ COLUMNS=150 dpkg -l apt | tail -n 1 [23:00] ii apt 0.7.25.3ubuntu9.3+ppa1 Advanced front-end for dpkg [23:02] wgrant: Is that the new apt? ^ [23:02] I presume so. [23:02] But looking. [23:03] Yes [23:03] :( [23:04] Hm? [23:04] Well, if it has the new apt and is still failing... [23:04] Oh, sorry. [23:04] That output is from the current build slave [23:04] As in, the test run that started ~ 30 minutes ago [23:05] Ah. I guess we'll see. [23:05] Thanks. [23:15] whats the ? to set webservice batch size ? [23:18] ?ws.size [23:18] thanks [23:19] hmm, I think I need more sample branches to fix properly. [23:19] shrug [23:20] Morning. Is there a URL to get the latest download release from a series? I'm writing a Makefile and would like to point wget at one URL that always resolves to the latest tarball. [23:20] (such as the latest from the 4.5 series at https://launchpad.net/gcc-linaro/+download) [23:22] I don't know [23:22] you could use the uwatch stuff that packages (like bzr) use to track upstream [23:23] wgrant: do you happen to know where the top level collections are glued into the web service? [23:23] lifeless: LaunchpadRootNavigation? [23:24] lifeless: OK, ta. I'll file a wishlist feature request [23:24] michaelh1: I'd suggest uwatch as the easiest way currently [23:25] indefinitely [23:25] lifeless: export_as_webservice_collection() makes it appear in launchpadlib, though. [23:25] realistically, such a link is going to be a long way away [23:25] we're sorting out many rather heavy lifting issues atm [23:25] wgrant: yes, I see. [23:25] wgrant: and then [23:25] *another* decorat [23:25] @collection_default_content [23:25] to select what is shown. [23:26] This is just so byzantine. [23:26] Right. [23:30] Ran 56 tests with 0 failures and 48 errors in 17.653 seconds. [23:30] "Bugger" [23:31] At least it was quick. [23:34] Total: 77 tests, 2 failures, 3 errors in 2 minutes 13.821 seconds. [23:34] Progress! [23:47] Is anyone else getting make errors (ImportError: No module named widgets) when they try and make on a new branch? [23:48] run utilities/update-sourcecode [23:49] wgrant: it would be nice for folk if when changing shipit you make a small song and dance on the dev list [23:59] lifeless: Thanks all working now.