[00:00] Just make sure lucid-lp-base isn't running while the clones are. [00:00] Doesn't matter wth the btrfs variant, but for aufs it does. [00:00] Also, it needs to have your SSH key in authorized_keys, as you may be able to tell. [00:03] There's probably some huge piece I've completely forgotten, but hunting for that can happen tomorrow. [00:09] Project db-devel build #676: FAILURE in 5 hr 48 min: https://lpci.wedontsleep.org/job/db-devel/676/ [00:09] wgrant: well, I am probably going to keep 100GB or so of windows around Just In Case [00:10] I'm assuming dmraid will handle the dell raid setup [00:10] Bah. [00:10] Possibly. === jkakar_ is now known as jkakar [03:21] mwhudson: you have a sandybridge machine right ? [03:38] lifeless: i do [03:50] did you have any issues with the natty desktop cd ? [03:50] I get the fine grey crosshatch and nothing more [03:53] lifeless: no, installation went absurdly smoothly [04:09] Yippie, build fixed! [04:09] Project devel build #846: FIXED in 5 hr 29 min: https://lpci.wedontsleep.org/job/devel/846/ [05:42] mwhudson: interestingly, its happy with lcd, but it hated my analogue monitor. [05:42] lifeless: heh [05:43] 50% of my desktop backed up [05:43] lifeless: i've tested mine with a projector over vga and that worked [05:43] (post-install) [05:43] mwhudson: this has a non-intel graphics card in it, for additional fun [05:43] being a desktop and all [05:44] ah ok [05:44] well i can't help you with that then :) [05:45] :) [06:00] Yippie, build fixed! [06:00] Project db-devel build #677: FIXED in 5 hr 51 min: https://lpci.wedontsleep.org/job/db-devel/677/ [06:26] jml: https://dev.launchpad.net/ArchitectureGuide/ServicesRequirements#preview === almaisan-away is now known as al-maisan === al-maisan is now known as almaisan-away [09:06] bigjools: Any objections to my comments on bug #797599? [09:07] <_mup_> Bug #797599: Copying archivepermissions cross-distro is wrong.

< https://launchpad.net/bugs/797599 > [09:07] wgrant: I don't see any comments from you [09:09] bigjools: Packagesets are per series. [09:09] bigjools: So after duplicating the sets we need to regrant permissions. [09:09] Why did that just get removed? It's trivial to fix. [09:09] launchpad_dev=# \d packageset [09:09] [...] [09:09] distroseries | integer | not null [09:12] abentley: You has QA. [09:12] wgrant: RARGH [09:12] Haha [09:12] That's my line. [09:12] bbiRargh? [09:12] Er. [09:12] bigjools: Rargh? [09:13] having to re-init *archive* level permissions on a new series is crack [09:13] Well. [09:13] the model is broken [09:13] Yes. [09:13] Sort of. [09:13] Packagesets have to be per-series. [09:13] I know [09:13] So permissions based on them have to be. [09:13] the FK is wrong [09:13] How? [09:13] Well, how would it be improved? [09:13] NFI ATM [09:13] Besides nuking from orbit. [09:13] Right. [09:14] Exactly. [09:14] need to think about it [09:14] but not in a plenary! [09:14] True. [09:14] I won't roll it back for now, but will file a critical regression bug. [09:14] that's fine, we can fix it [09:14] next week [09:16] morning d'blin [09:16] Hi lifeless. [09:17] apparently dmraid doth not split IO across underlying devices [09:17] this saddens me [09:17] Yay! [09:17] Fakeraid must die. [09:17] wgrant: Including MD? [09:18] wgrant: its not substantially different to mdraid other than needing to match BIOS supported layouts, *and* being able to boot sanely. [09:18] And wgrant has the deployment report open. Who would have thunk it. [09:18] wgrant: i like being able to boot sanely. [09:19] lifeless: Er? GRUB2 can boot off RAID [09:19] StevenK: last I saw it requires the device to be on a single disk, because its not actually bringing up all the layers [09:19] its essentially running in degraded form [09:20] lifeless: My fileserver boots off RAID1 just fine [09:20] this means it won't work for anything other than a vanilla mirror [09:20] But I can't check, since my home DSL is down. [09:20] StevenK: e.g. vanilla mirror [09:20] You're closer, so go check. :-P [09:20] StevenK: ring your wife :) [09:20] StevenK: MD is not fakeraid. [09:20] It's software RAID. [09:20] wgrant: so is fakeraid [09:21] It is, but with an extra layer of proprietary crap. [09:21] wgrant: its really not that proprietary. [09:21] crappy docs though [09:21] Stupid and pointless and non-standard, then. [09:21] wgrant: what would make it a standard? [09:21] Not being chipset-dependent. [09:21] It mainly exists because versions of Windows didn't do RAID, right? [09:22] StevenK: *booting* from RAID requires either fragile gymnastics or a BIOS (or lower) support for the raid layout [09:23] StevenK: the ICH (and similar) raid implementations exist because controller cards that have ATAPI/SCSI on the topside and N disks on the bottom side are hugely expensive for entry-level environments, and *other* than such cards ... you had to do gymnastics [09:24] so fakeraid does *exactly* what the early boot BIOS interrupt calls do for SCSI and IDE disks, and run just long enough to let the OS load its drivers. [09:25] if fakeraid is stupid, so is being able to boot off of IDE and SCSI disks. [09:30] wgrant: Is it possible to set append-revisions-only on an existing branch? I can only find references to it on bzr init. [09:31] gmb: You probably have to edit .bzr/branch/branch.conf using something like hitchhiker. [09:31] bzr reconfigure doesn't seem to have an option. [09:32] wgrant: Ah, right, thanks. [09:32] Just add a line with 'append_revisions_only = True' to branch.conf [09:37] You can use bzr config now to set append_revisions_only [09:37] Ahh, true. [09:37] Forgot about the new shiny. [09:43] LIFELESS! [09:43] RARGH [09:43] StevenK: Oh? [09:44] parellel-test #80 has been building for *13* hours [09:45] Heh. [09:46] hudson 25142 0.0 0.0 0 0 ? Z Jun28 0:14 [/usr/bin/python] [09:46] ORSUM [09:47] wgrant: Shall I kill it with fire? [09:54] jml: hi [09:54] jml: I've put some prose into the servicesrequirements page [09:54] lifeless: yeah, I saw, thanks. [09:54] lifeless: will look soon. [09:54] I think it captures all that we spoke about, and includes sufficient rationale [09:54] jml: if it doesn't, please fix-or-tell-me [09:54] ? [09:56] StevenK: heh, weren't you about to kill all of windmill with fire? :) [09:56] *windmill tests [09:57] Project parallel-test build #80: ABORTED in 12 hr: https://lpci.wedontsleep.org/job/parallel-test/80/ [09:57] DIE, FIEND [09:58] lol [09:58] lifeless: ok, will do. [09:59] nigelb: parallel-test has nothing to do with Windmill. [09:59] StevenK: ah! [09:59] One of my merge requests has become sort of stale. Before I work on that branch, I want to "rebase" it wth master [09:59] is there a bzr equivalent of that? [10:00] (this is an LP merge proporsal that I need to finish working on) [10:00] Don't rebase, just bzr merge devel in [10:01] lifeless: any ideas for useful LP (or related) hacking for me this week? [10:01] lifeless: :( resetting rabbit is hard. [10:02] jml: just about anything is useful [10:02] wgrant: Do you think there is any point to keeping the windmill-{db-,}devel jobs on Jenkins, or shall I delete them? [10:02] StevenK: I'd keep them for now, just in case. It is cheap... [10:02] jml: one thing that you and I probably have most experience around is getting the test suite to stop stomping on stdout when something goes wrong [10:02] lifeless: yeah, I know. that's largely the problem :D [10:02] jml: that messes up subunit [10:02] er, I did "bzr merge ../devel" and now it seems to have just gotten stuck [10:03] lifeless: yeah, ok, I could try with that. [10:03] nigelb: Is devel up-to-date? [10:03] doesn't merge give me feedback on stdout on what's going on? [10:03] StevenK: yeah [10:03] oh [10:03] done :D [10:03] Heh [10:03] I always wanted to make tests fail that spat stuff out to stdout or stderr [10:03] More patience required [10:03] Not sure the most bulletproof method of doing that though [10:03] stub: interestingly, the way we patch the login object practically guarantees that tests will [10:04] Sadly, I can't apt-get install patience :) [10:04] I'd like to see an option to bin/test that will stop spewing the librarian log out on test failure [10:04] deprecation warnings and bzr's __del__ methods are particularly hard to get at to fix this [10:04] StevenK: its an attachment to the test [10:04] StevenK: something you could do that is better [10:04] StevenK: would be to teach the layer to only attach the *per-test* component of the log [10:04] StevenK: rather than the log-since-fixture-start. [10:05] StevenK: e.g. truncate the log or record the position or whatever in the layer per test setup [10:05] StevenK: if you use tribunal though,t he librarian log won't be in your face anyway [10:05] lifeless: most of bzr's __del__ methods are being deleted. [10:05] spiv: I know :) [10:05] spiv: (and I'm glad) [10:05] lifeless: Bleh, more reason to learn tribunal [10:06] wow, I updated my devel again. [10:06] Lots of deletions \o/ [10:07] % apt-cache search tribunal | wc -l [10:07] 0 [10:07] I guess that's the windmill being burned with fire. [10:07] jml killed lots of line too [10:07] StevenK: poolie: can get you going with tribunal [10:07] I now realzie I'm going to fix the same bug 4 times. [10:08] <_mup_> Bug #4: Importing finished po doesn't change progressbar < https://launchpad.net/bugs/4 > [10:08] StevenK: it has testr integration, so its basically just tribunal . [10:08] I don't have testr either ... [10:08] StevenK: that one is packaged; shiniest version is in the testing cabal ppa, but the one in natty is eminently usable [10:11] lifeless: It seems there is no way to list all rabbitmq queues without using rabbitmqctl, which is 100-200ms :( [10:11] lifeless: Which makes fixture reset difficult. [10:13] wgrant: :( [10:13] wgrant: there are some alternatives [10:14] Creating a new vhost is also 100-200ms. [10:14] Seems to be erlang startup. [10:14] wgrant: such as, use the in python api to note queue access and connect-and-flush-all-those at the end of the test [10:15] lifeless: Yeah, possibly... [10:15] lifeless: But that becomes really bad when we have multiple consumers. [10:15] That we don't control. [10:15] wgrant: for an LP test run, all consumers should be test fakes [10:16] wgrant: which means we have a testing interface to them, and can tell them to do their own reset dance [10:16] "should" [10:16] wgrant: the page has been changed. [10:16] Also, I am talking about now. [10:16] Not in two years. [10:16] wgrant: *must*, or it can't go live. Existing things like the librarian are excluded [10:16] Hmm. [10:17] I'm flexible on this, but I think that being waffly would make it harder for folk to design well [10:17] so I've stopped waffling on this aspect. [10:19] by flexible I mean I am willing to be wrong and discuss; however your objection seemed to be that there was no requirement so we would have trouble in the future. [10:20] abentley: Can haz QA? And not submitting branches with [r=me] ? [10:20] Well, I told him to verbally, but I didn't think he'd take it literally :) [10:21] StevenK: do you mean 'please use r= rather than r=me when self reviewing' ? [10:21] StevenK: if so, +1 [10:22] r=me or rs=me is so pointless it may as well not be there. [10:22] Since loggerhead on devel won't tell us who commited it [10:23] Which means you either need to know, or ask buildbot. And asking buildbot makes me sad. [10:23] Or check the committer of the merge revs. [10:27] lifeless: Hmm, FixtureResource doesn't use reset()? [10:32] The rabbitmq management plugin does what we want, but the rabbitmq plugin build system seems to be designed primarily to make people want to die. [10:32] ha. headdesk. ha. [10:32] 103 files mention 'unseen' -- some would be false-positives [10:34] Perhaps we should use carrot. [10:35] (it has a memory-based backend, as well as a rabbit one) [10:36] Which would allow us to do unit tests with a nice fake. [10:54] wgrant: that only helps if we're not talking to a microservice [10:59] simple review needed https://code.launchpad.net/~jr/launchpad/797688-packaging-branches-label [11:04] Riddell: I'd like to see some test changes, to be honest. [11:06] someone asked about this yesterday and I wan't yet aware that it existed: http://ricostacruz.com/js2coffee/ [11:07] benji: wicked [11:10] poolie: this just came up on my feeds and wondered if you'd seen it: https://www.mergebox.net/ [11:10] they bill themselves as the GitHub of bzr [11:11] benji: they announced themselves on the bzr dev list [11:11] benji: very interesting project [11:12] ahh, I should have figured they would engauge in some way [11:12] not at all [11:12] its good to raise interesting things [11:12] lifeless: Sure. [11:23] Project parallel-test build #81: STILL FAILING in 1 hr 11 min: https://lpci.wedontsleep.org/job/parallel-test/81/ [11:33] usb2 is so slow! [11:33] bigjools: It is the 'ppa' name. [11:33] bigjools: Not 'Test' [11:33] So it's not quite so easy. [11:34] yeah just saw [11:51] Project db-devel build #678: FAILURE in 5 hr 51 min: https://lpci.wedontsleep.org/job/db-devel/678/ === almaisan-away is now known as al-maisan [12:01] Project parallel-test build #82: STILL FAILING in 37 min: https://lpci.wedontsleep.org/job/parallel-test/82/ [12:01] Launchpad Patch Queue Manager: [r=jml][no-qa] Remove syncUpdate. [12:07] benji, lifeless, has anyone actually got a login yet? [12:07] I haven't; I can't decide if I should, or should not :) [12:07] heh [12:12] poolie: maco did get a login for it if you're looking for feedback [12:12] (it being mergebox) [12:25] bigjools: dsds are shaping up really quite nice [12:25] lifeless: hellyeah! [12:25] bigjools: I hit a small usability thing today, not sure if its a bug or not. [12:26] bigjools: I wanted to answer, for a friend 'whats different between debian and ubuntu for virt-manager and libvirt-bin' [12:26] that is not enough info to properly ascertain an answer [12:27] bigjools: localpackagediffs doesn't seem to have a unfiltered view [12:27] it's always series-dependent [12:27] lifeless: ! [12:27] yes it does [12:27] like there is unignored, ignored, ignored-and-newer, and resolved [12:27] oh that [12:27] there's a bug about tha [12:27] t [12:27] kk [12:27] I shall not file one then [12:28] did I mention usb2 is painfully slow ? [12:28] lifeless: thx for feedback though [12:28] de nada [12:29] my friend uses Debian and then whinges that stuff doesn't Just Work :) [12:29] lifeless: USB 2 sucks, use something better. [12:29] there's an answer to that [12:29] StevenK: would-if-I-could [12:29] lifeless: There is this thing, called eSATA? [12:30] StevenK: external disk only having usb2 [12:30] StevenK: I just backed up everything from my old desktop onto it, and it got 30MB/s sustained... restoring at the same rate :( [12:32] lifeless: The last time I copied stuff over USB 2, I copied 800GiB at an average of 3MB/s, and it took 4 days. [12:34] StevenK: -ouch- [12:34] StevenK: I'm only doing 400GB and still its painfule [12:36] StevenK: https://launchpad.net/launchpad/+bugs?field.tag=easy+ui&field.tags_combinator=ALL === al-maisan is now known as almaisan-away [14:26] StevenK: I forgot to mention, if you fix a bug then tag it with "dublin" so we can keep track of the bugs fixed here === almaisan-away is now known as al-maisan [14:34] benji: I have updated this branch, just wanted to check the fix looks ok: https://code.launchpad.net/~huwshimi/launchpad/table-headings-728187/+merge/66008 [14:44] huwshimi: I made positive sounds on the MP. [14:44] benji: Thank you! === Ursinha_ is now known as Ursinha [15:00] StevenK: are you able to point me in the direction of writing tests for my 797688-packaging-branches-label branch at some point? === Riddell_ is now known as Riddell [15:56] benji: https://code.launchpad.net/~stevenk/launchpad/archive-picker-value/+merge/66318 [15:56] bigjools, StevenK, wgrant: any objections to a df app restart? [15:56] jtv: No, doit. [15:56] jtv: None. [15:56] StevenK: who are you calling a doit? [15:57] …or whommm? [15:57] jtv: You, duh. [15:57] oh, I see. [15:58] StevenK: done === danilo_ is now known as danilos [16:02] jtv, hey, I've got a question about the expander widget: how do I make it start out expanded? [16:03] danilos: disable JS. :) [16:03] jtv, ok, so that's a feature we are missing: if "expanded" flag is on it on load, we should probably not fold it [16:03] Sure. [16:03] jtv, it's a case I hit on a page [16:03] I mentioned this yesterday as "nice for later"; I guess "later" came early [16:03] jtv, ok, I'll assume that works and just set the "expanded" class on the content node [16:04] BTW where is that flag exactly? Is it a CSS class, a config arg..? [16:04] jtv, yeah, I was not sure if it was implemented or not, and it's needed to be able to replace the existing collapsible stuff [16:04] Ah [16:04] jtv, a CSS class state_marker.expanded [16:04] OK [16:07] StevenK: https://code.launchpad.net/~jr/launchpad/797688-packaging-branches-label/+merge/66146 [16:09] jtv: nup [16:10] ? [16:13] StevenK: voila [16:13] hmm, wait [16:31] benji: https://code.launchpad.net/~mbp/lazr.restfulclient/789369-getattr/+merge/62743 [16:41] anyone available to review a branch that attempts to fix bug 404279? https://code.launchpad.net/~matsubara/launchpad/404279-private-ppa-link/+merge/66335 [16:41] <_mup_> Bug #404279: apt urls on private ppa pages are incorrect (missing credentials)

< https://launchpad.net/bugs/404279 > [17:03] i will try [17:10] stub: addAfterCommitHook doesn't work.... [17:10] he no callee backee [17:10] You said it worked! [17:10] I LIED [17:11] commit is being called? [17:11] yes, we just stepped in pdb === al-maisan is now known as almaisan-away [17:12] bigjools: So we know the concept is sound, or else PostgreSQL connections would never be committed. [17:13] bigjools: Oh.... aftercommithook... not sure if they are used anywhere. [17:13] indeed [17:13] Maybe just a standard commit hook? [17:13] define "standard" ? [17:13] We can worry about ordering of rabbit or pg first later [17:13] the other one I can see in the doc is beforeCommitHook() [17:14] stub: maybe we need registerSynch [17:15] Now I can't find the documentation... [17:15] stub: http://readthedocs.org/docs/zodb/en/latest/transactions.html [17:15] It's not a synchronizer. [17:15] You can register a class with the prepare, commit, rollback hooks like a pg database connection does [17:15] It's not a beforeCommitHook. [17:15] It's also not an IDataManager [17:16] stub: how do you register that? [17:17] * stub wonders if our transaction package is out of date with those docs [17:17] and buggy [17:17] Writing it as a full blown datamanager should work [17:17] And not too traumatic since most of the methods will be noops [17:17] ok [17:18] stub: got an example we can see? [17:18] preferably in our code :) [17:19] bigjools: Those docs have one. Otherwise ZStorm has one. Nothing in the lp tree though. [17:20] The one in the docs will be simpler, and that api has been stable for over 10 years (apart from doom which we added) [17:23] Hmm... but I wonder if something else is going on. Given the aftercommit api exists, I would sort of expect it to do what it says on the tin [17:25] bigjools: Your hook has the right signature? [17:25] foo(success, args, kwargs) [17:25] kws [17:26] finished here - popping down [17:44] Yippie, build fixed! [17:44] Project db-devel build #679: FIXED in 5 hr 52 min: https://lpci.wedontsleep.org/job/db-devel/679/ === salgado is now known as salgado-lunch