[00:00] wallyworld: sweet, looks like I can remove the patch and just use the bootstrapped-ed-ed config for charmstore [00:01] wallyworld: stub calls check u have mentioned is done via CheckCall/ChaeckCallNames [00:58] hi wallyworld can i get ur 5 minutes on HO? [00:59] kelvin_: talking to vino, give m 5 [00:59] wallyworld, yup, thanks [01:06] wallyworld: I've pushed the test fix up as a non-squashed commit. You're a popular guy today though, I'll ask babbageclunk to take a quick look if he's free https://github.com/juju/juju/pull/8896/commits/eff1dc23ed0e400a21afa0d74e3a8c764e34ce39 [01:06] veebers: I'm free enough! [01:06] babbageclunk: awesome, cheers! [01:15] kelvin_: free now [01:15] veebers: yeah, "popular" [01:15] wallyworld, c u soon [01:25] veebers: that worked out nicely by the looks [01:26] was worth the time to avoid another patch of a global var [01:26] wallyworld: aye, wanted a sanity check on it to make sure the creation of that session wasn;t going to introduce some obscure issue [01:26] aye, just fixing another test (might get the same treatment) and should be able to land it [01:27] Sorry, got distracted - it looks nice to me too. [01:45] wallyworld, babbageclunk sweet, I did the same for UpgradeCharmStoreResourceSuite too [01:45] great ty [01:48] wallyworld: where are controller config options documented? For now I'll file a bug and assign it to me to follow up (for charmstore url) [01:48] there's a page in juju docs somewhere [01:48] not sure of exact url off hand [01:49] wallyworld: ack, I;ll file bug and keep track [01:49] ty [03:08] wallyworld: if a test does this: s.AddCall("NewCharmAdder", conn, bakeryClient, csURL, channel), can I do a check that "NewCharmAdder" was called, but I only care about one of the args? (i.e. csURL) [03:09] i think you can get specific args for a given call from the stub [03:10] and then assert on that [03:12] wallyworld: ack, cheers [03:26] wallyworld, thumper: What serialisation format do you think I should use for the FSM commands and snapshots? Might just make it yaml for now, it would be easy to change it (before release, I mean). Or gob would be another option. [03:29] i guess yaml is easier to debug? it's not like it will be huge quantities of data [03:40] wallyworld, babbageclunk: Whats the best way to debug this test failure :-( http://ci.jujucharms.com/job/github-check-merge-juju/2171/consoleFull (FAIL: machine_test.go:1021: MachineSuite.TestControllerModelWorkers) [03:40] introduced in a commit that I've since squashed up (yay) [03:41] veebers: havin' a look [03:41] cheers [03:41] veebers: loks like a build error [03:41] error: 820: package build failed: FAIL github.com/juju/juju/apiserver/facades/controller/charmrevisionupdater [03:42] wallyworld: I've sorted that issue out. This test fails locally too [03:42] do toy have a run without the build error? [03:43] wallyworld: not a complete one no. I can push up my fixes for the build error issue and get one though [03:43] ugh [03:44] veebers: that looks like a significant root cause because the test output says waiting for [charm-revision-updater] [03:44] and that's the thing that doesn't compile [03:44] wallyworld: ah, interesting. Ok I'll start looking there [03:45] also, there's a npe in the firewaller [03:45] and the panic in instancepoller sounds bad [03:45] the build error once fixed should address the worker bit of the failure [03:45] babbageclunk, wallyworld ah right, I'm seeing build issues I didn't see earlier. I have something to go on now, thanks! [03:46] good luck [03:47] veebers: Oh, but also the "[LOG] 0:18.065 DEBUG juju.worker.dependency "charm-revision-updater" manifold worker stopped: empty charmstore URL not valid" seems like it would be related to what you're doing. [03:47] babbageclunk: indeed! thanks [04:07] * vino going for lunch. [04:07] babbageclunk, wallyworld yay in the end an easy fix, had to set CharmStoreURL in the controllerConfigAttrs before the rest of the setup (the conns suite sets a default one which doesn't have that value) [04:08] gr8 [04:08] micro-gr8s [04:09] nice [04:42] wallyworld, r u free to talk about the bugs? [04:43] kelvin_: in a call, soon [04:44] wallyworld, yup, just ping me when u r ready, thanks. [05:32] wallyworld: ugh, my "microk8s.kubectl -n veebers delete po,svc --all" command has hung there for ages now :-\ seeing some DENIED in dmesg. Will explore after dinner [05:36] kelvinliu: finally free. jump in standup HO? [05:36] veebers: that happens to me sometimes too [05:36] i think i reinstall, not sure now [05:37] wallyworld, yup, [05:38] wallyworld: ack, will kill it shortly and try again [05:56] wallyworld: do we have a standard way to represent times and durations in yaml (so they can be roundtripped)? [05:56] hmmm. not sure [05:56] i know we tend to use fake times for that sort of thing [05:57] or we nil out times before asserting [05:57] we assrt that time value is not nil then make nil then assert [06:00] kelvin_: just a coouple of small comments [06:05] I mean, I can't do that because I need the time values for the leases. [06:18] babbageclunk: hmmm, i can't recall where we might do that [06:19] wallyworld: It looks like maybe it just does the right thing actually, just checking. [06:19] hope so === frankban|afk is now known as frankban [09:15] manadart: where we at with this, did you manage to test this locally? https://github.com/juju/juju/pull/8879 [09:38] stickupkid: I tested it out with the latest changes. Looks good to land and iterate on. [09:38] manadart: wanted to land this one before my others, so that's nice :D [09:39] stickupkid: I think all of yours are approved now. [09:46] manadart: perfect, thanks [10:11] should we be noting down which tests are flakey? [10:13] probably a question for rick_h when he comes online I guess [12:33] stickupkid: definitely, please file a bug for those [12:33] stickupkid: and see if you can find a tag that works for them [12:59] morning rick_h_, who do I poke for support on the new discourse.jujucharms.com? It doesn't seem to want to pull in my gravatar icon. Also are there some guidelines for what will/should be put there? Will existing things from the mailing lists and askubuntu be imported so it's actually consolidated? [13:00] zeestrat: otp, sec [13:28] all: how can i force juju to install lxd 3.0+ on xenial nodes? [13:34] rick_h_: Can you please help? [13:34] rathore_: howdy, otp with the team. sec [13:34] rick_h_: I am good thanks. Take your time [14:45] zeestrat: so, is 3.0 not the default in xenial after upgrade? Maybe it requires the backports PPA I guess. [14:45] or sorry, rathore_ that is [14:46] zeestrat: yea, not sure why the gravatar doesn't work. Please file a post in the issues section and I'll see if we can get the web folks to investigate. [14:46] zeestrat: as for porting stuff over, I'm not sure what we can do with the mailing list. [14:46] zeestrat: we are working to port over things like the Juju wiki pages, stuff in the juju/docs directory, etc. [14:47] zeestrat: and I don't think we'll pull askubuntu as that format kind of works there with the voting/etc. I'm not sure I'd want to pull all those over. [14:52] rick_h_: right, but it is supposed to replace the mailing lists? [14:52] zeestrat: yes, it is [14:52] zeestrat: not sure if we can put them in archive mode/etc. There's not an import/tooling to pull out the history I know of atm. [14:53] rick_h_: No still installs 2.xx [14:55] rick_h_: Gotha, thanks. I hope there's a way to pull the archives in some sort of way for searchability. Would hate to lose the history and context. [14:58] zeestrat: definitely, it'd be nice. There's scripts for importing user accounts from the email lists in mailing lists but not seeing anything for the topics. We'll see if anything comes up. [15:01] jam stickupkid: https://github.com/juju/juju/pull/8902 [15:20] rick_h_: Dumb question, is there a issues section/category on discourse that I'm not seeing? [15:21] zeestrat: oh hmm, thought there was. /me looks again [15:21] zeestrat: must have been cleaned out. I guess just go uncategorized please [15:26] rick_h_: Roger. [15:39] so it seems like it's the same issue over and over that's failing on for CI, I'm going to spend some time fixing it once I've finished my new PR, as it's currently a hindrance atm [15:43] stokachu: k, the action timeout thing? [15:43] stickupkid: that is [15:45] rick_h_: I believe it's in `juju/cmd/juju/application` - and it's around some mongo issue [15:46] rick_h_: the bigger issue is that we're hitting mongo at all, but that's another story... we're correcting as we go :) [15:47] stickupkid: always better than we found it and get there eventually :) [15:47] rick_h_: bingo! === frankban is now known as frankban|afk === freyes__ is now known as freyes [20:21] Morning all o/ [20:21] babbageclunk: ping [20:31] rick_h_: morning [20:31] thumper: morning [20:31] rick_h_: do you have a few minutes to chat? [20:31] I was waiting in the QA call [20:31] thumper: if that's not going down I do [20:31] chris declined [20:31] ah missed that [20:31] k, omw then to 1-1? [20:31] kk [20:33] rick_h_: I'm in the rick hangout because it looks like our 1:1 is missing a hangout/meet [20:33] thumper: I added it in, not sure what happend. [20:33] * rick_h_ must have not added it recurring last week [20:33] I see it now [20:33] wonder what happened [20:44] thumper: sorry, might have to swap the axis there to get it to show up right [20:44] brb, getting the boy [20:44] kk [21:11] kwmonroe: do we have this? https://twitter.com/esacteksab/status/1016428675930247174 [21:11] kwmonroe: as you mentioned a more concise changelog summary vs the 40 pages of stuff with the official release? [21:14] thumper: you still hacking on the graph? [21:14] no [21:14] I don't understand the underlying data, so hard to construct a query [21:15] thumper: ok, just wanted to make sure I wasn't stepping on your graph if I played with it [21:15] you're all good to play [21:21] hml: so, the feature flag worker [21:22] Where does your new worker run? [21:22] babbageclunk: in the machine agent [21:22] cool [21:23] That's the easiest option then [21:23] Ok, so you have a feature flag set up to determine whether your worker should run? [21:23] babbageclunk: yes [21:25] You can see the version I made for the raft enabled flag - you can just tell it the feature flag and whether it should invert the sense of the flag. [21:27] babbageclunk: for a normal FT, I don’t want to invert yes? [21:27] or correct [21:27] hml: yeah, that sounds right - the disable-raft one was a bit of a weird case. [21:28] babbageclunk: k - then i need the equivalent of ifRaftEnabled? [21:29] hml: yup - that enables you to wrap a given manifold so it only runs when the flag is true. [21:30] babbageclunk: diving in… :-) [21:31] babbageclunk: thank you! [21:32] hml: no worries - give me a yell if you have any more questions! [21:32] babbageclunk: :-) [21:40] would love a review in anyone is free https://github.com/juju/juju/pull/8905 [21:54] wallyworld: I'll take a look after my chat with thumper [21:55] you rock [21:55] ty [22:24] vino: hey, got a few minutes for a chat? [22:26] sure thumper. [22:26] 1:1? [22:26] ok.