[00:05] kelvinliu: does the charm supply the name for a mutating webhook config ? https://kubernetes.io/docs/reference/generated/kubernetes-api/v1.18/#mutatingwebhook-v1-admissionregistration-k8s-io [00:06] don't mean the object meta name [00:06] checking [00:09] yes it does [00:11] charm provides .meta.name and .WebHooks[].Name tlm [00:11] awesome thanks [00:11] np [02:38] wallyworld: thumper's not around, can you approve https://github.com/juju/juju/pull/11426 instead? (I'm sure he wouldn't mind) [02:39] ok, let me take a look [02:47] babbageclunk: just a couple of small nits [02:54] awesome, thanks === mup_ is now known as mup [03:08] kelvinliu: +1 on PR but maybe discuss? have a read of the comments [03:08] wallyworld: alright, I'm going to hit merge on that even though thumper's not approved it [03:09] he will get over it [03:11] wallyworld: I saw u mentioned more tests and to merge the hookChangeParams, HO? [03:11] kelvinliu: sure, give me 1 mnute to hit send on an email [03:12] yep [04:03] hey _thumper_ can you approve https://github.com/juju/juju/pull/11426? I tried getting wallyworld to approve it and then smash it in because you weren't around (having made the changes we talked about), but it looks like github prevents that. [04:03] <_thumper_> babbageclunk: sure [04:04] thanks! [04:04] <_thumper_> I did leave a comment before [04:05] <_thumper_> but didn't think I needed to approve as well [04:05] <_thumper_> but have done now [04:07] _thumper_: yeah, because you requested changes last time it enforces getting you to re-review [04:07] <_thumper_> yeah, sorry [04:07] <_thumper_> my "thumper" nick seems to be temporarily unavailable just now [04:07] <_thumper_> no idea why [04:08] he's been hacked! [04:08] <_thumper_> boo [04:29] _thumper_: glad this was caught. raised against 2.8-beta1 but affects 2.7.6 also https://github.com/juju/juju/pull/11427 [04:29] * _thumper_ looks [04:32] <_thumper_> wallyworld: approved, but you could simplify [04:32] ok [04:32] ty [04:33] _thumper_: you can but that's not why i got the controller state [04:33] i could have used either model or controller state to get config, but i had the controller one anyway [04:33] the controller state is used to get the cloud service info [04:34] <_thumper_> ah, ok [04:34] <_thumper_> is that controller specific? [04:34] not but here we are looking up the controller api addresses - the the controller cloud service mongo doc is what we need [04:35] as oopposed to a cloud service doc for a workload app [04:35] when we bootstrap, the controller get a k8s service mad efor it, like any other k8s app [04:36] and that service is mapped to a cloud service doc in mongo [04:36] and it records the address info [05:37] quick PR for anyone around https://github.com/juju/juju/pull/11428 [05:38] tlm: approved [05:38] cheers [05:55] manadart: would love a +1 on forward port of 2.7 when you're online https://github.com/juju/juju/pull/11429 [06:50] wallyworld: Looking now. [06:51] ta [07:01] wallyworld: you coming to this meeting too? [07:05] wallyworld: a tiny PR to add a helper func to caas layer for raw-k8s-set hook command, +1 plz https://github.com/juju-solutions/layer-caas-base/pull/26 [07:06] kelvinliu: ty, lgtm so long as it's been tested live [07:07] lol [07:08] could u just merge it plz ty [07:09] kelvinliu: merged, but mke sure to test to ensure there's no regression [07:12] sure tested by copying to charm, I will be testing it again many times tmr [07:12] all good :-) [07:55] wallyworld: You still there? [08:02] wallyworld: Nevermind; approved the merge patch. [08:28] wallyworld: https://github.com/juju/juju/pull/11395 is ready for re-review with unit tests and requested changes. Ping me tomorrow/tonight/tuesday if there are any issues. [08:35] hpidcock, JUST DON'T PING ME ON THE WEEKEND? [08:35] hah [09:12] hpidcock: ty, will likely look tomorrow [09:24] manadart, thoughts https://github.com/juju/juju/pull/11425 [09:32] manadart, wallyworld that forward port of 2.7 won't land as there is a compile error [10:29] ah balls, ok [10:35] stickupkid: manadart: it's because in 2.8 there's no uniter state file anymore, so the method tp load it has gone away from the source code, and the 2.7.6 upgrade step fails to compile, so i'll need to resurrect that code for the upgrade step [10:35] sick [10:35] i'll do it tomorrow [10:51] manadart, is this what you meant by endpoint bindings? Or are you talking about the charms endpoint bindings https://github.com/juju/juju/pull/11425 [11:00] stickupkid: What you have there is correct for now, but we also want to get st.AllEndpointBindings and continue if any of those are equal to the deletion candidate. [11:00] haha, don't delete anything in reality [11:02] manadart, also this is becoming somewhat a behemoth of a method, I may split it out [12:34] achilleasa: https://github.com/juju/juju/pull/11414 is ready for review. :-) [12:35] hml: will take a look in a few min; I have almost managed to bootstrap with an operator from my dockerhub account [12:35] :-D [13:22] achilleasa: looking at the unit test fails on 11414, looks like to do with no directory created. [13:29] manadart, what was the issue with contraints validation not being correctly implemented [13:30] stickupkid: It's naively correct for single network machines, but in order to handle multi-NIC, we need to compare units, not the application in aggregate. [13:30] ah [13:31] each unit lives on its own machine and guimaas is a good example that what spaces/etc are in play vary from machine to machine [13:45] achilleasa: wallyworld reminded me you can do a juju dumpdb too if you’ve enabled the develop feature flag at bootstrap. [13:45] instead of juju-db.bash [13:48] achilleasa: fixed fail in UniterSuite.TestUniterRelations, i *think* that’s the last one. [14:21] hml: any ideas how to upgrade on k8s? I am using dev mode but not sure how to get it to pull the new operator [14:22] achilleasa: ho? [14:22] sure [14:43] manadart, got a sec, I'm being dumb [14:44] stickupkid: In Daily. [14:49] achilleasa: we’re hitting https://bugs.launchpad.net/juju/+bug/1871496 [14:49] Bug #1871496: Juju fails on deploy with error cloud service not found for microk8s === narindergupta is now known as narinderguptamac [16:40] hml: left comments in 11414. I stil need to go through the last set of files but that probably means a second pass in the morning. Take a look esp at my last comments regarding the state implementation and let me know what you think [16:46] achilleasa: having the relation data SetState all at once, is definately a goal, but would be an even bigger rewrite of this code, i think. it sounds like polish to me? [16:48] hml: note that my comment was not to do it all at once. just have an object to call State once, then give you a method to get the state for a relation by ID and to save the state of a relation. You still make a SetState for one relation but avoid the Ensure call [16:49] the end result will be the same but the proposed way would likely involve less API calls and be a bit cleaner [16:51] also, having to rewrite a large chunk of code while in beta wouldn't really be considered as polishing ;-) [16:51] achilleasa: i’m not clear on why that avoids the Ensure call. The ensure call is made at other times, but it looks like the primary need is for new relations, not saved ones. [16:52] achilleasa: ho? [16:54] hml: omw [20:51] hml: you can do dump-db anytime if the feature flag is enabled, it's only a client side flag [20:51] wallyworld: ack [20:52] wallyworld: i referenced a sha for yo to fix the merge from 2.7 to develop [20:52] great ok, will look after next meeting, ty [20:57] wallyworld: hml's change looks good to me [20:59] yeah, was what i was going to do myself this morning, saved me the trouble :-) [21:00] wallyworld: it was blocking me for a bit… then other stuff came up after i came up with the fix. :-)