[00:52] perrito666: back [00:53] thumper: Updated lp:1645729 with profiling data now [01:05] axw: sorry persian king, I was having dinner [01:07] perrito666: np, let me know when you want to talk HA [01:09] now is a good time [01:16] axw: ? [01:16] perrito666: ok, HO? [01:16] perrito666: https://hangouts.google.com/hangouts/_/canonical.com/axw-perrito?authuser=1 [01:17] going [01:29] thumper: sorry, I was out yesterday when you pinged me. do you still have questions? [01:29] axw: yes... [01:29] I'm just making a coffee [01:30] I also have some heap profiles that I'd like to talk through with you [01:30] thumper: okey dokey. I shall make tea then [01:30] axw: now the x in your nickname makes me suspicious, are you actually named xerces? [01:30] axw: one with 25G [01:30] of heap [01:31] perrito666: hah no :) x stands for no middle name [01:31] I have 25G to discuss with you, that is a first :p [01:31] shoulda said yes [01:31] thumper: that's quite the heap [01:37] ha ha, Andrew Xerxes Wilkins would be quite the name. [01:47] * babbageclunk goes for a run [01:48] axw: let me know when you are ready to talk through this stuff [01:48] thumper: yep, in a few mins [01:50] thumper, axw, wallyworld: the automatic PR check the bot does seems to failing. it's the lxd tests. is that a known thing? [01:50] Failed to copy file. Source: /var/lib/jenkins/cloud-city/jes-homes/merge-juju-lxd/models/cache.yaml Destination: /var/lib/jenkins/workspace/github-check-merge-juju/artifacts/lxd/controller [01:50] not sure ... [01:50] not that i know of [01:51] maybe that's just a red herring [01:51] apparently the checks failed but that's the closest I see to a failure [01:51] so long as the merge happens, the pre checks can fail for now IMO [02:01] menn0, thumper I can be here a little bit longer, let me know if there is other info you need https://bugs.launchpad.net/juju/+bug/1646310 [02:01] Bug #1646310: Model Migration Fails on 3rd attempt === natefinch-afk is now known as natefinch [02:01] babbageclunk, ^^^ [02:01] alexisb: I'm talking with axw, will leave this for menn0 [02:02] alexisb: that's all the detail I need. thanks. [02:19] bradm: ping [02:20] * redir eods [02:38] alexisb: ok, looking [02:56] thumper: pong [03:01] menn0 are you chasing alexisb's bug? I'll stop looking if you are. [03:01] babbageclunk: i'm not at the moment was planning on it [03:04] wow, it suddenly turned into summer outside! [03:17] menn0: Ok, well, I've reproduced it - must be pretty new, because I was doing exactly this bouncing back and forth about a week ago. [03:17] babbageclunk: that's what I was thinking too [03:18] wallyworld: why are we copying this value to a temporary variable and then copying it back? https://github.com/juju/juju/blob/staging/cmd/juju/cloud/addcredential.go#L366 [03:19] natefinch: it's to do with the logic in the loop which handles fields with file attr set. i'd have to re-read the code [03:20] wallyworld: I guess I don't understand why we copy it, modify some fields, then copy it back, rather than just modifying those fields on the original [03:20] the currentAttr var is outside the loop [03:21] it's holds the value from the previous iteration, or something like that [03:21] and from memory, we need to know that for fields with file attr [03:22] we're copying currentAttr to fileAttr, modifying fileAttr, and then copying fileAttr BACK onto the original currentAttr. Why not just modify currentAttr directly? [03:22] axw: if you had a moment sometime, i'd appreciate a pre-impl check on this wip https://github.com/wallyworld/juju/compare/cmr-worker-publish-local...wallyworld:cmr-worker-publish-api?expand=1 [03:24] natefinch: i can't recall off hand anymore [03:24] well, I'll change it and see if the tests still pass :) [03:25] wallyworld: yup, knee deep in profiles atm, will take a look in a bit [03:27] axw: no worries, just when you are free. i'll keep progressing regardless as I can work on the facade backend. i'll also need a +1 on reed's review which the wip branches off sorry :-( but whenever is fine as I can keep occupied [03:43] Is there a charm that deploys as quickly as ubuntu used to? I think it's great that the new version of ubuntu uses lots more features, but it seems to take ages to deploy. [03:46] babbageclunk: just make your own, it's trivial [03:46] natefinch: yeah, good point [03:50] babbageclunk: here's my minimal charm. it's just a directory with this in it: [03:50] $ more min/metadata.yaml [03:50] name: min [03:50] summary: nope [03:50] description: nope [03:50] series: [03:50] - xenial [03:51] natefinch: awesome, copied! [03:51] I tried to drop summary and description but the juju yaml nazis yelled at me [04:50] wallyworld: do we actually have any credentials with optional values? [04:50] openstack - domain [04:50] that with keystone 3 [04:57] wallyworld: hmm ok [05:27] wallyworld: I thought we agreed that the worker would talk to its own controller always, and the controller would do the inter-controller connections [07:21] axw: ping? [07:43] babbageclunk: pong === frankban|afk is now known as frankban [08:16] mgz: you on yet? seen today's bot sadness? http://juju-ci.vapour.ws:8080/job/github-merge-juju/9754/artifact/artifacts/trusty-err.log/*view*/ [08:54] axw: oops - still around? [08:54] babbageclunk: yes, not for long though [08:55] babbageclunk: what's up? [08:55] axw: ok, just quickly - I just remembered the nasty patch hack we put in last time we were waiting for a fix to be merged into mgo [08:55] axw: do you know about that? [08:56] babbageclunk: I don't know specifics [08:56] babbageclunk: I think maybe mgz does? [08:56] babbageclunk: but I know what you're referring to, if that's enough :) [08:56] axw: yeah, he does as well [08:57] axw: yeah - you could create a patch there for your mgo change if you want to get it into the release [08:58] axw: it's a bit nasty but it'll solve the problem right now. [08:58] babbageclunk: yup, I think we'll need to do that for now. need to also update my PR with a test, which I expect will be difficult [08:59] so we can patch it in the meantime, since the change is pretty obvious [13:04] anyone got experience working with aws and cpu measurement? seems like cloudwatch is the only reliable tool, and that is kind of a bummer when using juju [13:05] from what i've read, when checking for example top, what is showing is the physical core rathe what you have allocated for your vm [13:05] so it's not going to give you accurate information if you want to monitor the resource usage [13:06] Bug #1571457 opened: Juju still vulnerable to CVE-2013-2566, CVE-2015-2808

[14:03] voidspace: is there any chance you could QA my 2.0 windows ssh branch? rick_h volunteered you :P === mup_ is now known as mup [14:13] im trying to bootstrap on rackspace but i'm seeing: error info: {"forbidden": {"message": "Policy doesn't allow compute_flavor:create:image_backed to be performed.", "code": 403}} [14:13] is rackspace still supported? [14:14] SimonKLB: definitely...not suer about that [14:14] SimonKLB: is this just straight from xenial? trying to do trusty or something? [14:15] simple `juju bootstrap rackspace` [14:15] * rick_h tries [14:15] SimonKLB: try with --debug and see if anything is fishy in there? [14:16] this might be something: juju.provider.openstack cinder.go:501 endpoint "volumev2" not found for "DFW" region, trying "volume" [14:19] mgz: does it require running Windoze? [14:19] mgz: I am happy to do that, it will take me some time as I do not have a windozen box [14:19] mgz: but I have license keys so I can create one [14:20] mgz: rick_h: ^^^^ [14:20] ah, /me forgot about that part [14:20] mgz: rick_h: I believe I still have an MSDN subscription as a python core developer [14:20] voidspace: nvm, this 2.0.2 > * [14:20] rick_h: kk [14:20] voidspace: just figured you might have time to help until hardware access came back, but if it's back nvm [14:20] mgz: sorry [14:20] we'll push natefinch for the help [14:21] rick_h: it is bootstrapping now, it wasn't earlier [14:21] SimonKLB: so took a little bit, but a bootstrap here just worked [14:21] SimonKLB: in the DFW region [14:21] voidspace: k, yay [14:21] hallo [14:21] rick_h: that's odd, could it be that i need a certain kind of user type? [14:21] natefinch: any chance of the QA review of mgz's PR up please? [14:21] rick_h: natefinch dissapproves of your choice of telegram for secure communication [14:21] SimonKLB: yea, I was just trying to look at what can be set on a user. [14:22] SimonKLB: because the error is a "not authorized 403" styl;e [14:22] rick_h: natefinch says it is insecure by default and we should use signal instead [14:22] also telegram rolled their own crypto... very big no no [14:22] voidspace: natefinch :P well don't send me your password in it but I've got it running with a couple dozen folks that I'm not interested in migrating one by one [14:22] * rick_h thinks folks should just use hangouts but wtf [14:23] for regular communication about BS work and home stuff, yeah, just use whatever [14:23] rick_h: you're the one who said we should install telegram for the sprint! [14:23] but sure [14:23] voidspace: yea, because that's where I've got other folks already [14:23] I had to be converted [14:23] I just like debate and I like to understand the state of the art [14:24] now you all suffer my lack of a spine to fight for anything else :P [14:24] ah, yea it's network effect [14:24] and I agree that encryption is probably not a super-high priority for work chat [14:24] natefinch: +1 for Signal [14:24] "team, let's meet for dinner in the lobby at 7pm" [14:24] oh noes! the NSA knows when to send the drone! [14:24] but do we want the NSA joining us [14:24] voidspace: maybe they'll pick up the bill? [14:25] even if the NSA aren't listening the googlebots are [14:25] to be fair, we're talking on a channel that is *publicly logged* right now. [14:26] * rick_h waits for ads to start showing up for signal [14:27] signal is 100% funded by donations and grants .... another plus for it. they're not *trying* to make money [14:27] sorry.. I just did a bunch of research on this stuff the day before yesterday [14:27] natefinch: rick_h voidspace just do as I did and come to live to nobodycaresaboutusland [14:28] lol [14:28] rick_h: found this https://community.rackspace.com/developers/f/7/t/5143 [14:28] perrito666: :-) [14:28] rick_h: "The problem appears to stem from the fact that you are trying to create a compute flavor which does not come with a boot disk. That error means that you have provided an image to use when building the server, but that the server does not have a boot disk." [14:28] perrito666: bring me treasures from your wonderful homeland to Barcelona [14:28] being boring is definitely your best defense [14:28] perrito666: probably waaaay too late for that now though [14:29] SimonKLB: reading [14:29] voidspace: actually I looked for your stone, but local hippies became too lazy and they all build the same cheap jewlery instead === jamespag` is now known as jamespage [14:32] perrito666: fair enough, no pressure dude [14:32] SimonKLB: ummm, I'm not sure what to do with that tbh. So Juju is handling this and it works here just as with the same command you ran. So should be same image/constraints and so no disk available? Some sort of timing issue? [14:33] SimonKLB: can you try another region? [14:33] perrito666: just find me incredible treasure some day, some time - that's all I ask ;-) [14:33] rick_h: sure, if you run it using --debug, do you also see: endpoint "volumev2" not found for "DFW" region, trying "volume" ? [14:33] else it might be that you have access to something that i don't [14:34] SimonKLB: retrying my bootstrap with --debug to see [14:34] thanks [14:36] SimonKLB: https://pastebin.canonical.com/172476/ nothing like that I can see in there [14:36] "You do not currently have access to the pastebin." [14:36] voidspace: you european folks always wanting to find treasures in latin america :p [14:36] ah sorry [14:36] * rick_h copies better [14:36] SimonKLB: http://paste.ubuntu.com/23563481/ take 2 [14:38] rick_h: looks like youre running some kind of local binary? [14:39] mine says: juju.environs.bootstrap tools.go:74 found 16 packaged agent binaries [14:39] SimonKLB: ah, true sorry. I forget to check out of the dev version [14:39] * rick_h goes back and tries again with the released version this time [14:42] perrito666: I want to find treasures everywhere [14:43] rick_h: I have to skip standup - minor family emergency [14:43] rick_h: nothing serious, difficult kid, tired wife [14:43] rick_h: you know where I'm at [14:44] rick_h: just finished deploy with vanilla 2.0.1 and the four bundles and *all four bundles worked* (cores allocated) [14:44] rick_h: that's new and different from before [14:44] rick_h: now tearing down and will repeat with 2.0.2 [14:44] rick_h: if that *doesn't* work then I finally have a real repro and something to work with [14:46] voidspace: rgr take care of the fam [14:47] 2.0.2 bootstrap in progress [14:47] hey, is anyone available for reviewing https://github.com/juju/juju/pull/6639 ? quick branch, thanks! [15:44] natefinch: so, I was mistaken about the ci status for add-credential, we have test coverage of autoload-credential but not add-credential [15:44] so I don't think your current branch with affect that. [15:55] katco: can you please lookg at frankban's branch above? ^ [15:56] rick_h: sec, talking with natefinch [15:56] katco: rgr ty [15:57] Bug #1646524 opened: Excessive logging in juju-db

[16:00] uiteam call now === frankban is now known as frankban|afk [16:26] frankban|afk: you have a review === frankban|afk is now known as frankban [17:01] katco: that was just a quick way forward. supporting constraints only as strings introduces a backward incompatibility in the API, and we already have config and configYAML in those same calls [17:01] katco: oh, and thanks for the review ;-) [17:02] frankban: we can version the facade can't we? [17:02] katco: yes we can, not sure is worth for this branch [17:04] frankban: we are already drowning in our technical debt, so i am going to take the stance that we need to do it correct the first time instead of continuing to paper over problems [17:04] frankban: a manager-type is free to override my judgment ;) [17:06] katco: ok, no problem, as I said, this was only an attempt. if we need to go for the non-shortcut solution, I'll try to find another bigger slot in the future, maybe after everyone is ok with the plan [17:07] frankban: sorry to be contrarian. we have just taken the "quick fix" route for so long it's now preventing us from doing new things [17:09] katco: no problem, I am no particularly attached to this solution or another, it just felt natural to me given we already have config and configYAML, didn't realize that's considered tech debt, but I can see the reasons [17:09] frankban: but specifically in a PR for this, i would be looking for supporting only one approach, and also to make a stab at true unit tests. [17:09] katco: why do you prefer stubs lately? time consuming tests? [17:09] frankban: the config is certainly precedent. i don't know whether it's good or bad :) [17:10] frankban: i have only been advocating for them lately, i have always preferred them for unit tests. here's why: http://martinfowler.com/bliki/TestPyramid.html [17:11] frankban: you don't *only* want stubs, but full-stack as the bottom of the pyramid is a death-sentence. we are experiencing that atm. things are too brittle; tests take something like 100-600% longer to write/fix than the code [17:12] frankban: this is also very good: https://testing.googleblog.com/2015/04/just-say-no-to-more-end-to-end-tests.html [17:13] katco: I'll read those, thanks! [17:13] frankban: ty for the discussion and open-mind. sorry again for the friction [18:16] brb errands === frankban is now known as frankban|afk [19:30] /me runs errands...bbiab [19:33] Bug #1646590 opened: Adding machine fails for Power8 node with permanent pending status [19:48] * redir really goes now [20:12] natefinch: looking at https://github.com/juju/juju/pull/6645 [20:12] babbageclunk: it's a WIP, so like, right now, there's no tests for a couple of the new functions I added [20:13] natefinch: oh ok - should I hold off? [20:13] babbageclunk: probably. I mean, it's totally functional, but I'll need to add a few tests [20:15] natefinch: duh, I totally blanked on the WIP at the start. [20:28] natefinch: or katco either of you up for some pair wtf'ing with me on a bug please? [20:28] rick_h: I love WTFnig [20:28] ing [20:28] natefinch: yay! a party it is then [20:29] natefinch: meet you in ?core when you're free [20:41] can someone +1 https://github.com/juju/juju/pull/6646? [20:43] and katco, re: tests; Amen! [20:43] balloons: :) [21:08] * redir back [21:09] morning [21:09] thumper: o/ [21:14] happy Friday to those celebrating it [21:16] I liked this: https://hypothes.is/blog/code-review-in-remote-teams/ [21:18] (In my experience our team's good at this.) [21:18] I mean, the Juju team as a whole, not the subteams. [21:30] babbageclunk, that is a great article, I haev not seen that before [21:32] alexisb: I like the idea that reviews should be good for team cohesion and morale - I've definitely worked in places where that wasn't true. [21:33] babbageclunk, agreed, also the bits on positive input and not doing the bare minimum [21:33] people put time into their code, reviews should reflect the effort [21:53] babbageclunk, is this bug still valid?: https://bugs.launchpad.net/juju/+bug/1620438 [21:53] Bug #1620438: model-migration pre-checks failed to catch issue: not idle

[21:58] alexisb: not sure - sounds related to something thumper was looking into? [22:00] alexisb: it sounds like the presence bug thumper was talking about can mean that the precheck could succeed when the interactive command runs, but then fail when the migration is actually starting. [22:00] maybe> [22:00] ? [22:01] yeah [22:01] but I fixed that [22:01] thumper, can you comment on teh bug please [22:01] * thumper goes to the bug [22:01] oh yeah [22:01] fixed that [22:05] thumper: PR that fixes it would be an awesome addition as a comment on the bug :D [22:06] :P [22:08] thank you thumper :) [22:09] rick_h, anastasiamac I need to change locations, may be just a few minutes late to our release call [22:11] brb, anothe rerrand [22:12] alexisb: k [23:32] jam: you're probably all over this, but: https://aws.amazon.com/blogs/aws/new-ipv6-support-for-ec2-instances-in-virtual-private-clouds/