[01:30] wallyworld: PTAL https://github.com/juju/juju/pull/10774 - update-cloud changes [01:42] https://github.com/juju/juju/pull/10763 updates for model log files on the controller [01:43] * wallyworld otp interviewing, soon [01:45] ack [01:49] thumper: approved [01:49] babbageclunk: thanks [01:59] hpidcock: got a sec? [02:00] yep [02:00] standup [02:08] thumper: super-simple review? https://github.com/juju/os/pull/14 [02:08] sure [02:09] babbageclunk: all good [02:09] to be followed by a dep update against 2.6 then a merge to 2.7 [02:24] thumper: looks like there was a backwards-incompatible change made to juju/os before mine (used in 2.7 but not 2.6). Should I just make the change to 2.7, or update 2.6 so that it can work without the removed function? [02:24] just looking to see how the juju/juju code was changed [02:42] hpidcock: i left a few more thoughts. see what you think, especiallyu about the worker setup [02:42] i need to head out to 1:1 in about 5 [02:45] wallyworld: have you got 2 mins then? [02:45] :)))) [02:46] bah looks like he already went [02:48] wallyworld: looks like an intermittent failure in apiserver/facades/client/action [02:48] thumper: can I grab you for a moment - confused about the series stuff [02:48] babbageclunk: sure [02:49] in 1:1 [02:49] i'll look after 1:1 when i get back [02:52] wallyworld: ack [02:57] wallyworld: problem appears to be iterating over an unsorted set, using Values() rather than SortedValues() in action.go method Tasks [03:04] nope, just saw the swap in the test... that's gross [03:07] wallyworld: I'm on it [03:15] wallyworld: I have the fix [03:15] if my branch fails to merge again, I'll add it to mine and retry [03:15] if it merges I'll propose separately [03:16] we have a 50/50 chance either way [03:16] map ordering FTW [03:17] it failed again, so pushing fix to my branch [03:17] https://github.com/juju/juju/pull/10763/commits/4d6fee29f7e7614f5d546220f3c83cd208abeed1 FWIW [03:21] I hate when a package has a directory that includes a broken symlink so they can test that it doesn't break them, and then it breaks any other tool that does stuff with that directory. [03:30] babbageclunk: like? [03:31] make rebuild-dependencies is failing for me because of gopkg.in/juju/charmrepo.v2/internal/test-charm-repo/series/format2/hooks/symlink [03:35] stink [03:37] it really is [03:40] just blowing away the vendor dir fixes it. [04:14] hpidcock: back from 1:1 now if you need me to look at anything [04:14] wallyworld: just getting the worker together and fixing tests I just broke [04:14] probably 30min away [04:14] no worries, i reckon we can squeeze it in [04:16] hah let's hope so [12:23] manadart: do either of 10772 and 10773 (the small refactoring ones) require additional qa to land? [12:25] hml: Probably prudent to flex them some. We can contrive some steps and discuss and stand-up. [12:25] *at stand-up. [12:26] manadart: agreed, just wasn’t sure if it’d been done already [13:42] rick_h: mind turning on github actions for juju/os? I can open PR with the workflow later [13:42] nammn_de: it's on [13:42] "enable local & third party actions for this repository" [13:42] nammn_de: they have to land first though [13:43] rick_h: ta, go it [13:43] *got it [14:37] simple review if someone is willing https://github.com/juju/juju/pull/10781 [14:51] achilleasa: ping [14:51] woooo! juju 2.7-beta available in the beta track now [14:51] errr 2.7-beta1 [14:51] hml: here [14:51] ty stickupkid!! [14:51] great stuff team! [14:51] * rick_h does a happy dance [14:51] stickupkid: hah, bug emails coming now [14:52] achilleasa: i’m not understanding WHY we validate the old default endpoint space when changing to a new one. [14:53] hml: we don't, we validate the new default if different than the old and not set to network.DefaultSpaceID [14:54] oh... I see what you mean. We can just always validate the new one right? [14:54] (so passing the old one through would not be needed) [14:55] achilleasa perhaps? i was misreading and not connecting that the oldDefaultSpaceIDForApp was only used to see if the value changes. [14:56] achilleasa: either way is okay, now that i get what’s going on. :-) [14:56] hml: you are right. Let me strip that out [15:00] hml: pushed a commit [15:00] achilleasa: ty, looking [15:03] achilleasa: nice, i think simplier easier to understand, and revalidating doesn’t hurt. [15:03] achilleasa: what are your thoughts around the Force values from the application and bind changes? [15:04] hml: did you see my comment in the other PR? [15:04] achilleasa: just need to do qa on 10780 [15:04] achilleasa: nope… will read now [15:04] hml: let me know if my reply makes sense :D [15:07] achilleasa: yes, ty. the corollary makes sense to me also. put that in the pr. [15:54] achilleasa: a question in 10780 before I approve. it’s logged in the pr [15:55] hml: I don't understand... This was caused by "spaceNotFeasibleError" which should work now [15:56] hml: it should have been fixed by this: https://github.com/juju/juju/pull/10780/commits/7bfbaa36128734ad63f0ed2c4d0bb7309f293e2c#diff-3b940d8ae0b2e8e93eabd0ea62b619feR314-R340 [15:56] achilleasa: hrm… let me try again. [15:57] hml: is that on guimaas? [15:57] achilleasa: yes [15:58] hml: I will try next with a few log bits in there to see what's going on [15:58] achilleasa: let me check something first. [15:59] achilleasa: shit… sorry. i think i know what’s going on. damn release. let me reproduce first [15:59] hml: I was about to suggest it's not running the right code :D [16:00] checked my history and i didn’t use —build-agent during bootstrap [16:00] probably picked up the wrong agents [16:13] stickupkid: os github actions and finally running on osx and windows https://github.com/juju/os/pull/15 :D [16:18] achilleasa: i had the wrong code. qa from scratch now === hardy is now known as falencastro [16:24] nammn_de, CR'd [16:27] stickupkid: thanks man. The reason I changed the signature of the function is that people in future working on it don't get confused as I do. Actually I was hell of confused what the reason for the test was and why it failed. As I am not on windows this was a not fun debug session :D. The function only get's used for the test anyway. [16:28] nammn_de, it gets used in newDetectHardwareScript in winrmprovisioner.go [16:28] nammn_de, and we would throw that info away in that function [16:28] stickupkid: ahh, we use it on other repos [16:29] nammn_de, yarp [16:29] stickupkid: then let me change it on test level. [16:29] nammn_de, hence why i think we shouldn't change the signature [16:29] wicked [16:29] :D [16:29] or WICKED [16:29] GREAT WORK [16:29] stickupkid: dammnit, this function itself is SIMPLE but CONFUSING at once :D [16:29] nammn_de, yeah, global state SUCKS [16:29] as for someone without context [16:29] nammn_de, no no, that package is horrid [16:30] stickupkid: ha, thats true 😿 [16:31] nammn_de, it should be take a inspector, which returns a host series as a dependency and then use the vistor pattern to walk over the series to ensure it's valid [16:31] nammn_de, that's what i'd do, but would cause pain to move to it [16:37] achilleasa: much better with the correct code! [16:43] stickupkid: yeah seems like we have a lot of places where we could add work. I guess takes quite some time no one is really willing to invest for some spaces :D [16:58] nammn_de, looks good to me [16:59] stickupkid: thanks [17:00] stickupkid: catalina support should already be merged, right? no more work to be done [17:00] ? [20:26] this is a really interesting post https://discourse.jujucharms.com/t/any-advice-on-using-juju-for-regular-websites/2223/6 [20:27] timClicks: yea, crazy post [20:28] single engineer has supported over 2,000 deployments (most test/staging, I expect) each with their own version of php, magento/wordpress, mail, SSO, ... [20:30] wheeeee [20:30] " [20:30] "part time" [20:41] timClicks: when you're free I'd love to catch up on how you'd like to go about the 2.7-beta1 release notes bit [20:42] timClicks: we didn't send anything out yet because of the debs we'll process tomorrow but wanted to sync on highest priority material. I tried to add some bits on the "to document" discourse post but I'm not sure if that's meant for beta1 release notes and such [20:42] ah right [20:43] I removed the non-snap instructions from the notes and pushed it love [20:43] *live [20:43] we can edit them back in once they're available [20:44] rick_h: did you want to jump into a call? [20:45] timClicks: sure thing, omw to the cross team [22:11] how is the lxd snap installer working with network restricted installs? [22:13] i.e does the apt proxy get passed to snapd? [22:15] admcleod: so there's a snap proxy config I would expect to be set [22:16] rick_h: in model-config? [22:16] admcleod: note this is 2.7 only as lxd is gotten from a deb before 2.7 [22:16] rick_h: riiiight ok [22:16] admcleod: and 2.7 has the updated code to handle lxd from a snap source and working in a better way [22:16] rick_h: ok thanks. wont log a bug then [22:23] rick_h: ad i seem to have wrked around it with 2.6.9 so \o/ [22:29] admcleod: woot [22:51] admcleod: would b awesome if u could document ur workaround in discourse \o/ well done!! [22:52] anastasiamac: i can do that if you tell me which one specifically [23:00] admcleod: I've create this for u using the wording in this conv... feel free to edit/comment [23:00] admcleod: tyvm [23:02] admcleod: https://discourse.jujucharms.com/t/pre-2-7-lxd-snap-installs-with-network-restrictions/2269 [23:05] anastasiamac: done [23:06] admcleod: \o/ [23:13] hpidcock: i have meetings for the next while soon, but i left a few comments to look at [23:14] thanks