=== Ursinha-afk is now known as Ursinha [01:26] mwhudson: ping [01:27] thumper: pong [01:27] mwhudson: remember that RT we had ages ago for a machine to test multiple codehosting front ends talking NFS? [01:28] mwhudson: feel like getting involved with the testing of said issue? [01:31] thumper: yeah, i'm subscribed to the RT [01:32] i wouldn't have a problem getting involved, i should probably ask linaro management if they do though :-) [01:36] mwhudson: ok [01:36] mwhudson: we want to test it but I don't have the resources to put people on it right now :-( [01:37] what do you estimate the investment to be? [01:37] and what do we get out of it? [01:37] we get better front end scalability for codehosting connections [01:37] and kill a single point of failure [01:38] not sure on time estimate [01:38] more tolerance of high load [01:38] I'm not sure on the time investment [01:38] I have a feeling that mwhudson would be able to give a better guess than me [01:38] less downtime [01:38] simple testing should be simple [01:38] [rephrased] [01:38] heh [01:38] if it all works fine, it's not long [01:39] although i don't know how one would test production like load safely [01:39] IS have provided the test rig, it's just hammering on it for a while? [01:39] yeah, & figuring out details of deployment i guess [01:40] sounds like something we can invest in, because I can certainly see the appeal [01:41] obviously if it starts to escalate we will have to look again [01:41] is jam's forking service stuff deployed yet? [01:42] I don't think so [01:42] mwhudson: I don't think it is being used yet [01:42] merged but unused [01:42] :( [01:42] somewhat tangential to the issue at hand of course [01:43] james_w: so you'd be ok with me spending tomorrow morning say on this? [01:43] mwhudson, sure. Let me know if it adds up to more than a day of your time. [01:44] tbh, i'm a little bit unsure what to work on as i have less than 2.5 days work left before being off for xmas, seems a bad time to start something big [01:44] yeah, this seems like a good stocking filler [01:44] though I might advocate for getting the dead code in use and only tackling one big infrastructure change at a time [01:45] I think jam's thing is stuck at RT 42199, the qastaging for it? [01:56] james_w: have you become a team lead ? [01:57] lifeless, yes, but things are still in flux [01:58] congrats [03:00] Anyone else looking at the devel->db-devel conflict? [03:01] hi lifeless [03:01] and stub [03:02] yo [03:02] thumper: are you around? [03:02] yeah [03:02] do any of you want to talk to me about bug 605775, and showing file content within loggerhead? [03:02] <_mup_> Bug #605775: Loggerhead doesn't support linking to the raw content < https://launchpad.net/bugs/605775 > [03:02] i'm wondering what fits best in lp [03:02] i guess i should go by analogy to bug attachments which is that we'll serve anything, but we must serve it from a separate domain [03:03] to avoid/reduce xss [03:06] poolie: I think that the option defined at the end of 2) to only support text/plain or application/octet-stream should mitigate the need for a separate domain [03:06] is ok [03:07] poolie: or at least have it a configurable option [03:07] as some people may want to run loggerhead locally and have it serve images [03:07] but launchpad doesn't [03:07] although... [03:07] if we did get around to adding wikkid into LP there would be similar problems [03:08] perhaps we don't care that much? [03:10] which "2"? [03:10] oh, i see [03:10] you have to pick from the list in my last post though :) [03:11] I was referring to the description [03:11] I haven't read the entire thread [03:11] is it necessary? [03:11] probably not [03:12] i kinda feel someone wearing a security hat ought to agree to it [03:12] agreed [03:12] depending on the size of the hat, you might get different answers :) [03:14] thumper: applicaiton-octet stream is insecure and must not be used the lp domains [03:14] (for user supplied content) [03:14] poolie: see... [03:14] s/the/in the/ [03:14] yes, i know [03:14] i'm trying to decide on the best alternative [03:14] I thought I put it in the bug already [03:15] I didn't read the entire thread [03:15] ah, I think my comments were in the merge proposal [03:15] you commented on the bug [03:16] https://code.launchpad.net/~mkanat/loggerhead/raw-controller/+merge/42675 [03:17] is where I linked to the documented ie behaviour [03:17] anyhoo, I'm not here today. ciao [03:17] so we agree on lp we will only serve them from a separate domain? [03:18] ciao [03:18] well [03:18] oh nm [03:19] showing raw content from private branches also sounds fun [03:19] probably needs a different hostname per branch or something :/ [03:19] Like we do for the restricted librarian. [03:20] yeah, and for the same reasons [03:20] The Web is bad. [03:20] :( [03:21] maybe we can use websockets somehow [03:21] (note: heavy sarcasm) [03:21] this all just makes me want to cry [03:21] if only people decided not to be evil [03:21] the world would be a nicer place [03:21] * thumper relocates back to his home [03:22] ok, we'll do the same as there [03:24] poolie: its a little complex to do right now [03:24] by do, I mean discuss [03:24] as mwhudson and wgrant are talking about [03:25] refusing to do it for private branches at all is an option too [03:25] poolie: my suggestion to mkanat to get a discussion going in the lp dev list remains as my best suggestion for moving forward [03:25] i agree too [03:25] I'm sure that if a proposal gets forged in that fire, it will be rock solid [03:25] We need an insecure domain. [03:25] I'm worried that landing an unsafe branch on trunk will make trunk un-upgradable to by LP [03:25] To stick stuff like the librarian and branches under. [03:26] wgrant: launchpadlibrarian.net :P [03:26] i am not going to let him land something that is unsafe by default [03:26] wgrant: the problem is not the insecure domain [03:26] coolio [03:26] even for non-lp users i think that would be a poor choice [03:26] have fun y'all [03:26] wgrant: the problems are the "slightly secure" domains [03:26] mwhudson: Right, but we sitll need an insecure domain. [03:27] let's confused the user and use lauchpad.net [03:27] Even mis-spelt? [03:29] thanks for your contribution [03:29] :P [04:29] thumper: Even if we restrict what mimetypes or extensions we serve from the Librarian, we want it on a separate domain because web browsers have bugs. [04:47] StevenK: wgrant: thee builder scores for PPA's - are they meaningless these days? or still relevant? istr some discussion a while back about no longer using those. ? [04:49] spm: They're still relevant. [04:50] We don't want to use them. [04:50] But there's nothing else yet. [04:50] We just can't stop [04:50] Right, because that [04:51] spm: Someone is asking for a bump? [04:58] well no. someone has a bump already - 12K and counting. but apparently not being built. [04:58] that's a built in bump. I've not touched a thing. [05:04] Hmm. [05:04] We are running low on buildds. [05:04] And there are some long builds running. [05:04] i'll do a patch for rlimits [05:05] spm i think robert would agree that if you have to kill/restart things by hand, there's at least two bugs [05:05] one underlying bug and one that it's not killed automaticalyl [05:06] launchpad is pretty hostile to the casual developer [05:07] poolie: agreed. I did like the idea of the feature flag - but accept the challenges there :-/ [05:07] i guess this isn't really news [05:24] spm do you have any idea what would be a reasonable limit? how much ram does that machine have? [05:24] shall we say 3G? or 2G? [05:25] 64K? [05:25] things were easier then [05:26] it has 8, but shares with some very big, and slow processes. it's not a single thing that gets bad per-se, it's the multiples that make it struggle. [05:26] I'd aim for 1.2->1.5, and see how often we abort. work from that?? (to open negotiations) [05:26] ok [05:27] this is another case where it would be great to set rlimits through flags so they could be tweaked [05:27] one step at a time [05:27] istr mwhudson did similar with the other whatsits, try and find a sweet point. [05:27] actually.... [05:27] I can have a look thru the last week and *give* you an idea of how much is normal. [05:27] doing new code rollouts seems like a tedious way to titrate it [05:28] right [05:28] wallyworld: wow, that's a strange test failure [05:28] mwhudson: yeah :-( [05:28] wallyworld: i think i know the problem though: for the tests in TestPullerMonitorProtocol, TestCase.setUp is being called twice [05:28] any ideas? [05:28] wallyworld: why this causes the symptoms it does.... no idea [05:28] oh [05:29] mwhudson: i'll just switch back to that branch and have a look [05:31] wallyworld: it's something like, the testcase's _recordOops method is registered as an @adapter(ErrorReportEvent) [05:32] wallyworld: and something about registering it twice means that it doesn't get deregistered properly [05:32] mwhudson: well, removing the dup setup worked \o/ [05:32] i was too busy looking for something non obvious and complicated :-) [05:33] thanks for finding it :-) [05:33] well, it was a good one for a set of fresh eyes i think [05:33] wallyworld: duplicating the line [05:33] self.useFixture(ZopeEventHandlerFixture(self._recordOops)) [05:33] in TestCase.setUp brings the error back [05:34] ok. interesting that it hasn't come up before now [05:35] yeah [05:35] * wallyworld fires up ec2 again :-) [05:38] i would like to know why the failure mode is just that strange [05:38] but i would also like to go home and have dinner [05:39] mwhudson: https://code.launchpad.net/~mbp/launchpad/690021-rlimit/+merge/43733 [05:39] but don't let it spoil/delay your dinner [05:39] mwhudson: enjoy your dinner. thanks again :-) [05:40] poolie: approved, you'll need to find someone else to land it though [05:40] (unless you can do that now?) [05:40] i probably have permission; i might need some hand holding [05:41] are there black box tests for this? [05:41] or, can i interactively test it? [05:41] * mwhudson points poolie at wallyworld :-) [05:41] poolie: yeah, there are blackbox tests [05:41] * wallyworld looks [05:44] poolie: i can help you land it but am not sure about testing it [05:44] i will see what i can do [05:47] * poolie fights with source code [06:00] Project db-devel build (217): FAILURE in 3 hr 34 min: https://hudson.wedontsleep.org/job/db-devel/217/ [06:02] * StevenK glares at Hudson [06:03] * wgrant glares back at StevenK. [06:03] You borked it! [06:03] Although I was watching :( [06:13] Project db-devel build (218): ABORTED in 12 min: https://hudson.wedontsleep.org/job/db-devel/218/ [06:13] Launchpad Patch Queue Manager: [rs=buildbot-poller] automatic merge from stable. Revisions: 12071, [06:13] 12072 included. [06:28] poolie: " I can have a look thru the last week and *give* you an idea of how much is normal." <== 99.55% are < 400Mb. 99.90% < 800Mb. then it jumps into crazy territory. I'd suggest starting with 600Mb. [06:44] spm, ok [06:44] so i put it up at 2gb [06:45] shrug that works. [06:48] === Top 10 Time Out Counts by Page ID === [06:48] Hard / Soft Page ID [06:48] 30 / 173 BugTask:+index [06:48] 10 / 112 ProjectGroupSet:CollectionResource:#project_groups [06:48] 9 / 2622 Archive:+index [06:48] 9 / 230 Distribution:+bugs [06:48] 7 / 235 Distribution:+bugtarget-portlet-bugfilters-stats [06:48] 6 / 16 Archive:+copy-packages [06:48] 6 / 7 Person:+bugs [06:48] 5 / 195 POFile:+translate [06:48] 5 / 6 ProjectGroup:+milestones [06:48] 4 / 0 https://api.edge.launchpad.net [06:58] 'morning [06:59] heya jelmer! === almaisan-away is now known as al-maisan [07:30] wgrant: bug 690484 is a dup of #? [07:30] <_mup_> Bug #690484: "Start in 1 minute" shown for hours despite 12505 points < https://launchpad.net/bugs/690484 > [07:32] lifeless: It may not be a dupe. [07:32] The underlying problem was not fixable until very recently. [07:33] lifeless: there's a basic summary in #is from about 90 mins ago. fwiw. [07:34] wgrant: I thought there was a bug for queue estimating already [07:34] oh, I see [07:34] PEBCAK [07:35] It's not really an estimate issue. [07:35] right. it's not an estimate problem. [07:35] It's publisher latency + people clicking things that they shouldn't. [07:35] so, someone will updates it? [07:35] if anything the estimate was accurate. [07:35] I will talk to Julian about it this evening. [07:35] just... wrong. :-) [07:35] Since it's now feasible to fix. [07:36] wgrant: so the estimate would be "Never. This PPA is unable to publish" or words similar? [07:36] spm: No -- private PPAs need no longer wait for publication. [07:36] ah. even better. [07:37] They can use the public restricted librarian. [07:44] hy bryce [08:03] how do i actually land something through pqm? [08:03] i don't need ec2 testing [08:04] bzr lp-land, or bzr pqm-submit. [08:04] lp-land takes an MP, pqm-submit requires you to specify the source, destination and commit message manually. [08:05] ok [08:07] bryceh: so my current theory as i said in mail is that for some tests there are two database connections present [08:07] and two transactions [08:08] ok [08:08] this may be bs [08:09] i was hoping someone would answer that it was true :) [08:11] poolie, http://pastebin.com/JFiz4mga [08:11] wait, nevermind that [08:11] (ran the wrong test) [08:13] iow i'm wondering if the test is running in a different storm context from the code under test [08:13] or something like that [08:13] here we go - http://pastebin.com/UghMbtsJ [08:13] bear in mind it's only installed per-thread [08:14] note down towards the bottom Features: {'bugtracker_components': None} [08:14] I assume that should show as something other than None once the feature flag is seen [08:14] right [08:15] that does show that at least your template/view checked the flag [08:15] in -1 seconds! [08:15] way to go [08:15] hehe [08:17] i see that create_view does setFirstLayer etc [08:17] i wonder if that gives it a new db connection? [08:17] or maybe this is a different kind of layer? [08:19] I tried doing this, but same error: [08:19] Store.of(self.bug_tracker).add(FeatureFlag( [08:19] scope=u'default', flag=u'bugtracker_components', [08:19] value=u'on', priority=1)) [08:19] hm [08:19] actually if my theory is true you'd expect that no uncommitted changes are seen by the view [08:19] but in fact it's very common that they all are [08:20] can you make your view print its threadid and print the tid from the test code? [08:21] you mean like: [08:21] print view.threadid [08:21] print self.tid [08:21] ? [08:21] no i think [08:22] threading.currentThread() [08:26] _MainThread(MainThread, started -1219778880) [08:28] the same in both? [08:34] hmm, I can't figure out how to get the view to print its threadid [08:35] ok [08:35] bryceh: i need to go and cook [08:35] the next thing i would try is to get all the feature control rules [08:36] and see if there is one in there that's not being bmatched [08:40] ok good night all [08:40] bryceh: if you learn anything else (or don't), please send a followup to the dev list [08:40] poolie, ok thanks for trying [08:41] oh you could also search the archives [08:41] this feature flag stuff has been a huge headache [08:41] somebody, maybe deryck, hit something similar before [08:41] sorry :( [08:41] for testing, or generally? [08:41] generally [08:42] if you can send a mail explaining where the problems have been i'd appreciate it [08:42] was hard to figure out how to get it set up and working in the first place, now its hard to figure out how to get it to work in tests [08:42] poolie, that's easy - the documentation is really confusing [08:43] ok, that should be fixable then [08:43] anything else? [08:44] well, if you think of anything, let me know [08:52] wallyworld_: ping [08:52] thumper: hi. [08:52] wallyworld_: https://code.qastaging.launchpad.net/+daily-builds [08:52] wallyworld_: can you land a patch rs=me to remove the distro from the first column? [08:52] i know. the filtering doesn't work :-( [08:52] wallyworld_: just the distribution fmt:link [08:52] i'm fixing it now [08:53] wallyworld_: this isn't the filtering [08:53] i'll include your change in the same branch [08:53] wallyworld_: thanks [08:53] yeah, i was just letting you know i was already looking at it [08:53] i'll do it tonight and the mp will be in your inbox tonmorrow morning [08:54] thumper: i assume the distro is to be removed because it's always Ubuntu? [08:54] wallyworld_: yep [08:54] wallyworld_: perhaps in the future if it isn't always ubuntu, we may change it [08:55] thumper: i was just typing that question :-) [08:55] :) [08:56] thumper: mw found the codehosting issue too :-) [08:56] wallyworld_: awesome [08:56] mwhudson: I owe you a drink next time we are together in the same place :) [08:56] thumper: there was a bad test that called setup twice - been there a while, unrelated to my changes. not sure why it showed up the way it did [08:56] oh well [08:56] good to clean it up [08:57] yep. it's in ec2 again as we speak. it was such a simple error - i was looking for something really complicated [08:57] :) [08:57] * thumper afk again [08:57] wood, tree and all that [09:11] Hello [09:29] Does anybody have any objections to a reversion of r12041 (the Bugs restricted LFA API change, which has been qa-bad and untouched for four days)? [09:33] wgrant: ISTR abel doing some work on that, but he's got a bad back and won't be working for the rest of the week. Probably best to check with deryck, who I think knows what's happening there. [09:34] gmb: k, was wondering where he was. [09:34] Thanks. [09:34] Will check with deryck when he appears. [09:35] Cool. [09:46] has anybody else tried ec2 land in natty yet? From the first look of it it looks broken here. [09:49] I was going to upgrade this weekend. Maybe I should not. [10:14] Anyone know off the top of their heads how I can persuade a widget to sprout a tabindex? Or should I just poke around until Zope blows up in my face, as usually happens with widgety things. [10:17] Nice. I'm so smart I ask the question and then /part the channel. I'm obviously going to have a winner of a day. [10:20] * gmb decides to just go an shove things in templates. It works better. [10:21] As long as you don't use BeautifulSoup... [10:24] Do we have reports of script runtimes somewhere? [10:25] Ah, BeautifulSoup, the thing that made syncing with Sourceforge possible for a short while. [10:26] IIRC it was also involved in generation of the code import registration form. [10:26] wgrant: IDK; mthaddon might know. There may be something at https://lpstats.canonical.com. [10:27] we don't - it's in the DB, but I don't believe we graph it [10:27] (scriptactivity table) [10:28] Hmm. [10:46] Yippie, build fixed! [10:46] Project db-devel build (219): FIXED in 3 hr 35 min: https://hudson.wedontsleep.org/job/db-devel/219/ [10:46] gmb: Do you know what the deal is with revision 12041 on https://devpad.canonical.com/~lpqateam/qa_reports/deployment-stable.html? [10:49] allenap`: wgrant already wondered about that. I don't know where abel was up to with fixing it. deryck might; if it's okay with him we'll back it out. [10:50] gmb: It turns out that bigjools managed to get a rev after it rolled out anyway, so I am no longer strongly desiring its removal. [10:51] wgrant: Righto. Worth pursuing anyway, elsewise it'll make things confusing. [10:51] there was a rev that backed it out, no? === matsubara-afk is now known as matsubara [11:41] bigjools: I sure hope so, or apport is broke again :) [11:46] It was only rolled out to Soyuz. [11:47] It's not been rolled back, AFAICT. [12:04] Morning, all. === mrevell is now known as mrevell-lunch [12:56] Is the thing that generates diffs for merge proposals single threaded? No update for 8 mins on https://code.launchpad.net/~stub/launchpad/pending-db-changes/+merge/43763 === mrevell-lunch is now known as mrevell [13:29] fwiw that MP still hasn't been updated - and it's 39 minutes ago now === flacoste changed the topic of #launchpad-dev to: Launchpad Development Channel | BUG JAM! http://mumak.net/lp-bugjam-2010/ 35 so far | New starter this week: wgrant | PQM open for business | firefighting: - | Get the code: https://dev.launchpad.net/Getting === Ursinha is now known as Ursinha-lunch === matsubara is now known as matsubara-lunch [14:59] abentley, adeuring, allenap , bac, benji, danilo, sinzui, deryck, EdwinGrubbs, flacoste, gary, gmb, henninge, jelmer, jcsackett, jtv, bigjools, leonard, mars, mrevell: Reviewers Meeting ping [15:00] ack, thanks for the ping [15:00] * gary_poster is ready with my "me" [15:02] allenap`: ^^ === allenap` is now known as allenap [15:03] Thanks bac. === Ursinha-lunch is now known as Ursinha === matsubara-lunch is now known as matsubara === Ursinha is now known as Ursinha-afk === Ursinha-afk is now known as Ursinha === leonardr is now known as leonardr-afk === beuno is now known as beuno-lunch [16:37] Hey sinzui, do you have time for our call now? [16:37] mrevell, I will be free in a few minutes [16:37] * sinzui otp [16:37] np [16:42] gary_poster: Do you know if it's possible to give a widget or action button a tabindex using Zope? (Sorry if that's repeated; I lost conn) [16:42] gary_poster: I'm asking in the context of https://bugs.launchpad.net/malone/+bug/558419 [16:42] <_mup_> Bug #558419: +filebug page really should have tabindex for "Submit Bug Report" Button

< https://launchpad.net/bugs/558419 > [16:43] gmb: I'm pretty sure it is one of the knobs. Lemme see if see any docs... [16:44] mrevell, mumble [16:44] gary_poster: Thanks. [16:49] gmb: "extra" is a string that is just dumped in the widget tag, I believe. you can set it up with the usual customwidget bits [16:49] is that enough to get you going, or should I try for more detail? :-) [16:49] gary_poster: Brilliant, thanks for the tip. [16:50] cool [16:50] gary_poster: That's fine. I'll poke around til something works - or breaks. [16:50] :-) === beuno-lunch is now known as beuno [17:44] flacoste: so I am not sure after that call as to whether jml's bugjam counter is looking for the tag or not, what was the consensus? [17:45] bigjools: sinzui thinks it is, i'm not sure [17:45] bigjools: looking at the code would clear doubts [17:45] but i'm not sure where the code lives [17:45] it would. but ... exactly [17:45] putting the tag on is safe anyway [17:45] where is it running [17:45] and good practices [17:45] on his server i think [17:45] it's on mumak.net [17:46] an API script [17:46] yeah, I just went through 20-odd bugs adding the tag [17:46] let's see if the count changes [17:46] bigjools: it doesn't to seem to run that often either [17:47] but we'll see [17:50] and on that note, it's good night from me === benji is now known as benji-lunch === al-maisan is now known as almaisan-away [18:52] abentley, hai! I wants lp-propose! where do I get it from? [18:53] beuno: it's in the launchpad plugin [18:53] so if you have bzr you should have itg [18:53] s/itg/it/ [18:53] orly? [18:54] zomg! [18:54] that is awesome [18:54] thanks jelmer === benji-lunch is now known as benji [19:07] gmb: hoping to have a bugs team review on a revert I'm doing of an adeuring branch. You happen to be around for such a thing? I think you are the only candidate === Ursinha is now known as Ursinha-brb === matsubara is now known as matsubara-afk [21:09]