[00:06] babbageclunk, et al.: have any time to take a look at https://github.com/juju/juju/pull/10461 before standup? [00:11] timClicks: sorry, just looking briefly now [02:18] babbageclunk: re the vm template's name [02:19] perhaps it could simply be "juju-unit-template"? [02:19] because it will be hosted in a directory that's unique to each controller [02:20] or "juju-{series}-template"? [02:20] we could use the SHA256 string, but that feels unnecessarily complicated [02:21] timClicks: I know in the vmdk code it had some extra structure to disambiguate... just looking hang on [02:21] I think series was one part, was there anything else? [02:21] the series is still there, fwiw - was part of the directory [02:22] used to be /{controller uuid}/{series}/{sha265}.vmdk I think [02:22] PR is currently using /{controller uuid}/{series}/juju-{model uuid}-template [02:23] Ah, right - I think the rationale for the hash was so that you'd get a new image if it was released. But is that right? Wouldn't it just use it if it was already there? [02:23] timClicks: jump into standup? [02:23] ok [03:16] babbageclunk: can you test the internal vsphere instance? it doesn't to be responsive [03:19] timClicks: trying [03:20] timClicks: works for me (tm) [03:20] maybe vpn? [03:56] babbageclunk: hrm, it appears that turning things on and off again isn't helping [03:58] nm [03:58] looks like changing to the us-based VPN is getting me through [04:58] babbageclunk: hey final thought on template naming [04:58] is there any need for it to be namespaced by the controller uuid? [05:09] timClicks: sorry, missed this! Dumb mouse ran out of batteries [05:09] timClicks: if it's not by controller ID then we don't know when to clean it up. [05:09] that's a good point [05:10] (without implementing some kind of reference counting but I don't know how we'd do that) [11:21] can I please get a CR for https://github.com/juju/juju/pull/10484? [11:29] achilleasa: I am having a look. [11:34] achilleasa, man, they don't sort the bundle changes in pylib [11:34] achilleasa, :| [11:35] stickupkid: does it even work for more complicated bundles? [11:36] achilleasa, i have no idea, probably not [11:36] stickupkid: I remember we were hitting issues with relations not being created in the correct order when we were working on the bundle cmr stuff [11:36] achilleasa, guess where i'm at now :D [11:36] haha [11:36] aha! [11:37] achilleasa, i really don't want to copy the same sort method, but I guess I have to? [11:37] achilleasa, i.e. i don't like it's non-terminal if things go wrong [11:38] achilleasa, anyway, i'm going grabbing some food, so i'll think over lunch [11:38] stickupkid: yeah... I wonder if you can somehow make it bail out if it exceeds a large number of comparisons (of course, finding a safe value for that is definitely not trivial) [12:10] manadart: can you take a look at this one too? https://github.com/juju/charm/pull/289 (this one is for fixing the relative bundle issues in the develop branch) [12:13] achilleasa: Sure. [12:30] achilleasa: Approved. [12:30] tyvm [16:11] can I get a CR for https://github.com/juju/juju/pull/10487? (it's the new and improved fix for resolving relative charm paths - for dev) [16:12] rick_h: this one ^^ also works with overlays [16:12] achilleasa: yay [16:12] achilleasa, i'll have a look soon, fighting with circular dependencies of deps :( [16:42] achilleasa, approved [16:47] stickupkid: tyvm [17:19] rick_h, so that worked https://github.com/juju/juju/pull/10471 - we just need to fork this into juju namespace and we're good to go https://github.com/SimonRichardson/rpcreflect