[00:46] wallyworld: the review tool change is not ready yet. `confinement 'classic' not allowed with plugs/slots lint-snap-v2_confinement_classic_with_interfaces` [01:22] kelvinliu: ah, they are behind schedule. i'll email them to ask [01:22] wallyworld: ok thx [08:25] Need a forward merge review: https://github.com/juju/juju/pull/10978 [09:46] manadart stickupkid as you guys already gave it a quick look, its ready for a review. Mostly little cleanup and updated the go doc https://github.com/juju/juju/pull/10967 [09:47] btw. manadart: there is no way for a controller to access another controllers collection, right? [09:47] at least not without a lot of overhead (?) [09:47] nammn_de1, nope [09:47] stickupkid: thanksss [09:47] nammn_de1, you'd need to write a LOT of api server/client apis [09:49] stickupkid: is there a quick way to find out from the package naming which runs on a controller and which on the units on the not controller? Or something similiar? [09:51] nammn_de1, https://github.com/juju/juju/tree/develop/apiserver/facades [09:52] stickupkid: I see, somehow overlooked this 😭 [09:52] Interesting to hear about the CentOS + VMware issue, I know some people that are struggling with that, or at least would like to use it. I've used MAAS + my own CentOS images to be able to use Juju with CentOS until I figured out how to do it with LXD instead, but VMware would provide more use cases [12:13] manadart: this is the final instancepoller PR https://github.com/juju/juju/pull/10979. I have also noted my findings about the dups in the linklayer devices/addresses collections [12:13] achilleasa: OK; will look in a bit. [12:14] manadart: take your time... it's a big one :-( === tinwood_ is now known as tinwood [15:07] manadart did you mean that by renaming in your pr review? https://github.com/juju/juju/blob/4ad19fc6fc3105d0d94404cedfd9be33db4f34dc/state/modelmigration.go#L861 Wasn;t 100% sure [15:07] wanted to make sure before merging [15:20] rick_h: what would be the ideal output of this bug be? https://bugs.launchpad.net/juju/+bug/1853529 [15:20] possible options: https://pastebin.canonical.com/p/fVh89BcHQ9/ [15:20] Bug #1853529: The lack of a controller is not an error [16:07] manadart, when you get 5 minutes https://github.com/juju/description/pull/67 [16:07] i'm going to start on external controller [16:13] achilleasa: Still NFI what is up with that. This fixes the error https://pastebin.canonical.com/p/kqVJGHFjJS/ but the test still times out waiting for the address. [16:15] nammn_de1: I think option 3, just "No controllers registered" [16:16] rick_h: but this would mean for all the other commands then as well? E.g. [16:16] juju offers, juju status, juju models are all returning the same message. Do we want to change them as well? [16:24] nammn_de1: Yes, and rename LatestRemovedModelMigration as well. That can call CompletedMigrationForModel. [17:42] manadart, external controllers done, I'm now going to wire them up, before landing the PRs https://github.com/juju/description/pull/68 [17:43] manadart, I'm unsure how relation id plays into this whole thing, so I may end up removing it before landing #67 [17:45] manadart: that's odd... I will spend some time in the morn commenting out things from the linked commit to see if I can find the block that makes it fail in the first place === timClicks__ is now known as timClicks [20:24] timClicks__: did you get the provider instance types answer yesterday? [20:33] hml: not really, but I don't think that we have a lot of capacity available to implement what erik/scania would like [20:34] which is supporting centos7 images on vsphere [20:41] timClicks: I only know a few, and not how it works for vsphere. [20:41] timClicks: with vsphere it’s in the config file we setup i believe. [20:41] timClicks: the other is aws, were we pull down the data, and parse it for our use. usually check when we update AZ [20:42] yeah, we build a custom base template image [20:42] timClicks: rackspace would be the same as openstack [20:42] sure [20:43] timClicks: at a very high level, if the vsphere has centos7, why couldn’t we use it in the config file? [20:44] I believe we could, which is why I'm probing around asking questions. [20:46] timClicks: :-D [20:49] timClicks: related to the bug, the providers do add model config keys, at leaste the openstack provider does. does vsphere not add the ones in your bug? [20:50] no, it doesn't [20:50] or at least, this is what users see when they try to set up vsphere [20:50] timClicks: hrm.. the values are there… but only if you have bootstrapped vsphere. [20:51] timClicks: https://github.com/juju/juju/blob/develop/provider/vsphere/config.go#L17 [20:52] timClicks: juju show-cloud --include-config [20:53] wonder why they aren’t seen? [20:56] i don’t have a vsphere cloud defined locally to check. [20:58] hml: timClicks what do we need? I've got one now :) [20:58] rick_h: I just filed a bug [20:58] rick_h: run the juju command above please [20:58] timClicks: were you bootstraped to vsphere when you ran the commands? [20:59] i didn't, I copied that output from a discourse post from someone else [20:59] hml: timClicks ok, ran the command...what am I looking for? [20:59] rick_h: what does “The available config options specific to vsphere clouds are:” [20:59] list [21:00] oh bummer I don't get to do that since I'm on JAAS [21:00] timClicks: okay, i’ll check disposable2 [21:00] so it doesn't give me cloud config in that way [21:00] rick_h: hrm, not cool, can i get a pastebin of what you do see? [21:01] I meant I’ll check discourse