[07:21] cjwatson: Found any issues for RTM on DF? [07:22] wgrant: Nothing yet. Getting livefs builds working was a bit exciting, but not really for any reasons connected to Launchpad. [07:45] wgrant: Just set lp.distributions["ubuntu-rtm"].development_series_alias to "devel" to match Ubuntu [07:46] Might possibly be worth having that as part of the init procedure [07:57] cjwatson: Noted. [09:39] hi, I keep getting a timeout error when doing: w3m -dump https://launchpad.net/ubuntu/+archivemirrors-rss (latest oops-id is OOPS-c3c091162163ac61751c1a003d5c8fdf). is there a workaround or anything so that I can get a cached copy? update-manager needs this to update its internal mirror list [09:46] Hmm [09:47] I thought I increased that timeout to like 20s last release. [09:47] Surely it can't be timing out still... [09:47] Oh, that was +cdmirrors-rss. [09:47] I'll have a look at archivemirrors after dinner. [09:47] mvo_: ^^ [10:09] wgrant: Shall I do the LFC DB patch now, or are you on it? [10:10] stub: I haven't started yet, so feel free. [10:10] ta [10:10] Thanks. [10:10] It's not terribly complicated, hopefully. [10:10] Pretty simple except for the pg_attribute hackery. [10:12] stub: While you're there, might be worth making LFC.sha256 NOT NULL as well. The only reason it's not already is that I didn't care enough to hack pg_attribute, but we have to for the PK so we might as well kill two birds. [10:13] wgrant: Should I just reuse the sha256 as the PK? [10:13] Bah. stupid. [10:13] stub: No, that makes it impossible to split. [10:17] Have you timed just using ALTER TABLE for the NOT NULLs? I'd rather avoid hacking the system catalog if I can. [10:17] I tried that with a table that I forget and ended up with three minutes of downtime, even when I thought it was very hot. [10:18] And LFC is not small. [10:18] It's even reasonably wide due to the hashes. [10:18] From http://www.postgresql.org/docs/9.3/static/catalog-pg-attribute.html: [10:18] "This represents a not-null constraint. It is possible to change this column to enable or disable the constraint." [10:18] That sounds like an invitation to me. [10:24] thanks a lot wgrant [10:48] mvo_: "fixed" [10:59] wgrant: Is it possible to create non-Ubuntu PPAs yet? I see that createPPA doesn't take a distribution= kwarg [11:01] cjwatson: dogfood has two unlanded branches merged: overrides and createPPA(distribution=) [11:02] Ah, I should look at DF's apidoc [11:04] Oh, argh, PPA names are unique regardless of distribution, aren't they? [11:05] They shouldn't be, but I probably haven't hacked createPPA hard enough, fixing. [11:05] I still need to do some rearrangements on haetae before they can be ununique on prod, though. [11:06] And I have deleted https://dogfood.paddev.net/~ci-train-ppa-service/+archive/ubuntu/landing-000 due to abject confusion, bah [11:06] Heh. [11:06] How do I go about undeleting that? I guess I need to run something to fully process the deletion? [11:07] Undelete or recreate, I suppose. [11:07] I'd just set the status back to 0. [11:07] But the PPA publisher will set it to DELETED and rename it out of the way so you can recreate it. [11:07] Oh right [11:07] I'll flip the status [11:09] ok, undeleted, thanks [11:10] let me know when I can try createPPA again [11:13] cjwatson: Fixed. [11:13] Thanks, looks good. Error message in the case of a clash could possibly be improved now. [11:14] I wonder if we should put the distribution name in the rendering of PPAs on Person:+index. [11:14] https://dogfood.paddev.net/~ci-train-ppa-service looks somewhat confusing. [11:15] I've thought about that on occasion. [11:15] My most favourable proposal is to just split them up into sections by distribution. [11:16] Mm, yeah. [11:16] That's nicer than "(Ubuntu)" x a zillion [11:16] Also just pulled a fix not obliterate all similarly named PPAs when you delete one of them. [11:16] Hah [11:17] similarly> same name different distro? [11:17] Right. [11:17] Deletion killed /wgrant/ppa rather than /wgrant/ppa/ubuntu, for good reason (.htaccess, .htpasswd live in /wgrant/ppa rather than /wgrant/ppa/ubuntu) [11:18] But haetae's Apache config is fixed, so I can push that all down a level now. [11:18] The ArchiveDependency UI should be terribly broken, but I think non-Ubuntu PPAs otherwise work. [11:19] We'll find out soon, I imagine, as sil2100 is working on getting citrain working [11:20] Oh, there's no non-Ubuntu PPA publisher cronned atm, should fix that. [11:20] In fact I don't think there's a PPA publisher cronned at all? [11:21] I don't think so, no. [11:33] cjwatson: Both are publishing at */1 now. I've set publish=false on most Ubuntu PPAs so it should be quick. [11:36] Cool, thanks. [11:46] cjwatson: Do we want something like the archive reference as well, perhaps? [11:46] I don't know how the ddeb fetcher works, and I guess it's not worth making it too much more reliable at this point. [11:55] wgrant: pitti tells me that it otherwise works; in fact he's ignoring Distribution right now and deriving from Release, but he probably shouldn't be [11:56] Translations might care though. [11:57] cjwatson: The translations sbuild hack hasn't been using in very nearly eight years [11:57] ... [11:57] used [11:57] I suspect it's not even triggered any more. [11:57] Ah, good [11:57] So it's just to make ddebs a little less horrible, and because --archive=ubuntu in the config file is disgusting [11:58] Quite. [11:58] (That option being called --archive is disgusting, too, but) [12:04] cjwatson: I'm initialising some more distros on DF for QA, so don't expect the derived primary publisher to be particularly responsive at all times. [12:09] wgrant: ta [13:25] Hm [13:26] Creating a full copy of precise amd64 and i386 with packagecopier doesn't work. [13:26] CannotCopy(u'libunwind 0.99-0.3ubuntu1 in precise (binaries conflicting with the existing ones)',) [13:26] I wonder why. [13:26] Perhaps the primary archive is inconsistent.