[00:57] wallyworld: plz comment on bug 1672879 when u get a chance [00:57] Bug #1672879: [2.1.1] Controller does not look at agent_stream in model-config - it uses agent_stream that it booted with

[01:09] wallyworld: can I get a review for my fix for the GCE legacy network bug? https://github.com/juju/juju/pull/7150 [01:14] babbageclunk: sure [01:14] wallyworld: fanks === thumper is now known as thumper-headdown [01:16] axw: ping [01:16] menn0: pong [01:21] axw: regarding the rev menioned in bug 1669180, that's quite an old rev for a recent ticket. how sure are you that it's the right one? [01:21] Bug #1669180: proxy-ssh/juju ssh --proxy is ignored [01:22] axw: it seems like jam's more recent work is what Gabriel/you are referring to [01:22] * axw looks [01:22] babbageclunk: minor nits only [01:23] axw: it does look like jam may have done something weird with proxy handling in his recent changes though [01:23] menn0: hmm yeah doesn't look like it was changed in that rev, sorry [01:23] axw: ok cool... just making sure i'm not going crazy [01:23] menn0: oh wait [01:23] it is [01:24] * axw gets the line [01:24] menn0: https://github.com/juju/juju/commit/d25d100f3c04eb6dc22c58db51e87c1d947a6836#diff-f75a59a4d41206aca31bb545d00e3e31R326 [01:25] axw: ha, that's the line I was looking at too. [01:25] axw: for some reason I had thought that was part of jam's recent changes [01:25] axw: I agree that it's fishy [01:26] axw: thanks i'll dig further [01:26] menn0: no worries [01:39] wallyworld: I called those network initially but that clashed with the network package I use just further down. Do you think I should rename the import? [01:40] babbageclunk: nah, tis ok [01:40] leave as is tehn [01:40] wallyworld: I agreee that it's a bit awkward though. [01:41] wallyworld: just checking against a non-legacy network to be sure. [01:43] ty [02:09] anyone... https://github.com/juju/description/pull/6 [02:09] juju branch to follow [02:26] thumper-headdown: i'll swap ur PR for thoughts on bug 1675048 :) way forward or solution would b nice \o/ [02:26] Bug #1675048: mongod doesn't return when killed [02:26] * anastasiamac looking now [02:40] menn0: your change seems sane to me [02:40] axw: did you see my email? [02:40] oh I see email, /me reads [02:40] axw: just more detail [02:41] axw: I'd like jam's feedback before proceeding [02:41] axw: we might end up going with this for 2.2 and something more sophisticated for 2.3 [02:42] menn0: I suppose an alternative to classifying addresses would be to find an address that's routable from the controller's perspective. that might be as simple as finding addresses in the same space [02:42] menn0: but yeah, I think this is probably fine for now [02:43] and now I see I basically just said the same thing as your last sentence, which I hadn't got to yet ;:p [02:57] ah fuck... [02:58] * thumper-headdown headdesks gently [03:01] hmm... [03:01] maybe not === thumper-headdown is now known as thumper [03:19] well shit [03:19] axw added model status back in april 2016 and migration missed it [03:19] and here I thought it was something new [03:20] it was new in april 2016 :) [03:20] not helpful [03:20] :( [03:20] on the plus side, can make it required in the description rather than optional [03:21] which is cleaner [03:21] it seems to me that we need to have better testing that description pkg is in sync with what we have in juju.... [03:27] well... this was missed when it was together, so that isn't the issue [03:28] exactly.. what ele could have been missed in the same manner?.. [03:28] else* [03:36] wallyworld: I'm trying to test the add-subnet part of my provider-network-id change, but I can't run it with GCE because add-subnet requires a space name and spaces aren't supported on GCE. Should I add a card to make add-subnet accept an empty space name? [03:46] wallyworld: PR for storing network id: https://github.com/juju/juju/pull/7152 [03:48] babbageclunk: looking [03:49] wallyworld: ta [03:49] babbageclunk: empty space name is what we want, tes [03:50] wallyworld: I have pods running in kubernetes in vsphere [03:50] I'll do some more testing, but looking good [03:50] whoot! [03:50] proxy did the trick? [03:51] wallyworld: yep [03:58] babbageclunk: my OCD kicked in with naming consistency, see what you thnik [03:58] wallyworld: the problem is there are already inconsistent names, so you're never going to be happy. :( [03:59] true. i was hoping not to make it worse :-) [03:59] but, your call [03:59] wallyworld: This is definitely better - there are heaps of other ones on network.InterfaceInfo that are all in this direction. [04:00] righto [04:01] wallyworld: So, I could make them consistent with the other names in the same structure, but then we're going to have weird inconsistencies between the types. Maybe that's better? [04:02] hmmm, maybe best to stick with something that's more consistent overall [04:02] lessen the tech debt [04:02] wallyworld: Yeah, that was where I ended up, I think. [04:03] sgtm then [04:03] wallyworld: sorry, kind of annoying. [04:03] yeah, can't be helped, it's what was there [04:03] wallyworld: I might quickly get add-subnet working with no space, then? [04:04] why not [04:06] wallyworld: since it's a positional arg (with optional availability zones after it) I think it needs to be required as "" though. Too crufty? [04:07] oooh, yuk [04:07] but what other choice is there that retains compatibility [04:07] maybe we add a --space arg [04:08] and deprecate the positional arg [04:08] or if --space is used, don't interpret the first psoitional arg as a space [04:08] just assume ones [04:08] zones [04:08] so we support old syntax but add better syntax [04:09] Ok, so it would be: if there's a --space option, the positional args are cidr-or-provider-id [zone1 zone2...] [04:11] if more than one positional arg and no --space, the 2nd arg is space name (preserving backwards compat) [04:11] and we allow just 1 arg (and --space not specified) meaning add this subnet not in a space? [04:12] I think that works. [04:14] Has the possibility of putting something into a space accidentally if someone has spaces named the same as an AZ and they don't know about the legacy mode? That would be pretty weird though. [04:15] wallyworld: ^? [04:16] babbageclunk: yeah +1 but i think 1 arg should error as it does now [04:16] but the error would mention the prefered syntax [04:17] so the preferred syntax for adding a subnet without a space would be `add-subnet --space=""` ? Seems janky. [04:17] oops, `add-subnet --space=""` [04:17] wallyworld: ^ [04:19] wallyworld: confirmed, kubernetes is happy now. ran the microbot smoke test, and it's responding happily [04:19] * axw moves on to the vswitch bug [04:19] babbageclunk: or add-subnet --space foo mysubnet [04:20] axw: quick review? https://github.com/juju/names/pull/79 [04:20] looking [04:20] axw: awesome about k8 [04:20] wallyworld: right, but the no-space version would be add-subnet --space "" mysubnet? [04:21] babbageclunk: yeah :-( maybe just keep as is [04:21] and can be bikeshedded with folks next week [04:21] babbageclunk: or propose and we can ask for input on the pr [04:23] I'll send an email to jam and cc you - think I should add anyone else on? [04:24] wallyworld: ^ (I'm still trying to get the hang of when to put someone's nick at the start of messages) [04:24] babbageclunk: sorry, i keep switching back to IDE [04:24] include witold as well since he's in that space (pun intended) [04:25] wallyworld: ha, good friday afternoon joke! [04:25] dad joke [04:58] babbageclunk: you're adding subnets to the "undefined" space, thus "" is perfectly appropriate for now [04:58] babbageclunk: that's concretely the "what do we want to call the otherwise-unnamed space" [04:58] right now it is exactly called "" [08:45] jam: ok, thanks - so I understand that as: we still require the space to be passed positionally, but we allow "". Right? [08:47] jam: I guess in general, we'd want people to be using named spaces. So this is enough of a special case that it's ok for for the command to do it is a little awkward? [08:48] gah, edito: ".... for the command to do it to be a little awkward?" === admcleod is now known as admcleod_afk [14:49] wallyworld: I've updated https://bugs.launchpad.net/juju/+bug/1314682 with what I'm getting [14:49] Bug #1314682: Bootstrap fails, missing /var/lib/juju/nonce.txt (containing 'user-admin:bootstrap')

[15:06] sinzui: ^^^ [15:08] wpk: did you setup the dns/addresses to be ip4 [15:14] wpk: did the instructions in comment 28 not work? [15:38] sinzui: cloud-init finishes, apt-get update takes ~1.5 seconds [15:39] there is an error in cloud-init.log but found non-fatal and unrelated === akhavr1 is now known as akhavr === akhavr1 is now known as akhavr