[00:08] <menn0> wallyworld: I've been consistenly getting these for a while on the develop branch: http://paste.ubuntu.com/25266422/
[00:08] <menn0> just rebased and I'm still getting them
[00:09] <wallyworld> interesting ok. haven't seen those
[00:09] <wallyworld> oh wait
[00:09] <wallyworld> that i think is a mongo bug
[00:10] <wallyworld> we needed to upgrade to 3.2 to solve it
[00:10] <wallyworld> it happened on 2.4 in CI
[00:10] <wallyworld> but since using 3.2 it hasn't been seen since
[00:10] <menn0> wallyworld: that makes sense. the machine i generally run tests on has 2.4
[00:11] <menn0> wallyworld: we still need to support 2.4 b/c of trusty
[00:11] <wallyworld> sure, but if there's a mongo bug....
[00:11] <menn0> wallyworld: what's the bug?
[00:11] <wallyworld> i can't recall now, i'd have to google it
[00:12] <menn0> wallyworld, axw, babbageclunk: big model refactoring (part 1/2): https://github.com/juju/juju/pull/7710
[00:12] <menn0> wallyworld: never mind
[00:12] <menn0> wallyworld: about the bug, not the review
[00:12] <wallyworld> :-)
[00:13] <wallyworld> menn0: "which meant that the Model referred to a state.State associated with the correct model."  <----- that should be "incorrect" right?
[00:13] <menn0> wallyworld, axw, babbageclunk: and just remember it was 10x more painful for me to make these changes than it will be for you to review them :)
[00:13] <menn0> wallyworld: yes!
[00:13]  * menn0 fixes
[00:19] <babbageclunk> menn0: Oh wow, sounds like a fun change
[00:20] <wallyworld> menn0: i wish we didn't have *both* Tag() and UUID() methods at the sate layer - should be UUID() only. tags are for the apiserver layers
[00:20] <menn0> wallyworld: I completely agree but they've got their little taggy claws in everywhere now
[00:21] <wallyworld> yeah i know :-(
[00:21] <wallyworld> just having a bitch
[00:21] <wallyworld> i once had a gf like that
[00:23] <wallyworld> menn0: although in places in this PR the UUID() variant could easily be used instead
[00:23] <menn0> wallyworld: yes there are
[00:23] <menn0> wallyworld: I had to draw a line somewhere
[00:23] <wallyworld> that is true
[00:24] <menn0> wallyworld: if I have the energy after getting rid of State.GetModel, I'll try and tidy that up too
[00:24]  * menn0 makes note
[00:24] <wallyworld> sgtm
[00:24] <wallyworld> tbh, getting rid of ForModel() makes me very happy
[00:24] <menn0> me too
[00:24] <menn0> I wasn't even aiming to do that but it just fell out
[00:30] <wallyworld> menn0: just going to standup, left an initial comment
[00:31] <menn0> wallyworld: ok thanks
[01:29] <thumper> axw: quick question for manual provisioning, the user on the host needs to have sudo permissions (I'm assuming) to configure the service, but does it have to be passwordless sudo?
[01:31] <axw> thumper: no, the process will set up passwordless if it's not done already: https://github.com/juju/juju/blob/develop/environs/manual/sshprovisioner/sshprovisioner.go#L31
[01:31] <thumper> sweet
[01:31] <thumper> ta
[01:32] <wallyworld> menn0: quick chat? easier than typing into a pr
[01:32] <axw> babbageclunk: FYI, you'll need to use a proxy for external access on that MAAS: http://squid.internal:3128
[01:32] <axw> babbageclunk: otherwise LXC containers won't start
[01:32] <babbageclunk> axw: ooh, thanks.
[01:32] <axw> babbageclunk: and if you didn't already find out, it's ppc64el-only
[01:34] <babbageclunk> axw: Ahh, just saw your reply about the staging key - thanks
[01:34] <axw> np
[02:11] <menn0> wallyworld: now ok?
[02:13] <wallyworld> menn0: sure, griday ho?
[02:14] <menn0> wallyworld: ok
[02:25] <thumper> OH FFS
[02:25] <thumper> every time I'm about to start writing this spec, I end up replying to another email
[02:26] <thumper> now I get my physio reminder
[02:26] <thumper> I feel I'll be back in the evening to work on this spec, and disconnect email
[02:37] <babbageclunk> griday ho
[02:41] <anastasiamac> babbageclunk: glad u said it :D
[02:58] <axw> babbageclunk: in case you missed it, PTAL at https://github.com/juju/1.25-upgrade/pull/13 when you have some time
[02:58] <babbageclunk> axw: sorry, I did miss it - will go through it in a bit.
[02:59] <axw> babbageclunk: aw crap I just realised, this isn't going to work on the MAAS thumper provided, because it's ppc64el... the upgrade command assumes the client and server are the same arch
[02:59] <axw> le sigh
[03:00]  * axw wonders if the controllers we intend to upgrade are amd64
[03:00] <babbageclunk> axw: oh yeah, that's annoying.
[03:19] <thumper> babbageclunk: griday?
[03:20] <babbageclunk> thumper: sure
[03:20] <thumper> I don't get it
[03:20] <babbageclunk> 14:13 <wallyworld> menn0: sure, griday ho?
[03:21] <wallyworld> sigh
[03:21] <anastasiamac> :D
[03:21] <wallyworld> "g" and "f" are next to each other
[03:21] <wallyworld> you all know what i meant
[03:21] <thumper> oh...
[03:21] <anastasiamac> comic relief is nice this windy tuesday :D and yes, we did and that only helped the laughter
[03:22] <thumper> I didn't notice that wallyworld said it first
[03:22] <babbageclunk> It was just enough typos and jargon that it tickled me.
[03:22] <thumper> I thought it was a reference to me...
[03:22] <babbageclunk> It can be.
[03:42] <wallyworld> veebers: is the landing bot down?
[03:42] <veebers> wallyworld: let me have a look, you have a link to what you're trying to land?
[03:44] <wallyworld> veebers: https://github.com/juju/juju/pull/7708
[03:45] <veebers> I'll take a look
[03:46] <wallyworld> ta
[03:54] <veebers> wallyworld: that should be unstuck now
[03:54] <wallyworld> aesome ty
[03:55] <wallyworld> so it is
[04:01] <bradm> is there anything I can look at to see why a juju status is taking a long time?  this stack is taking anywhere between 7 and 9 minutes just on a bare juju status
[04:18] <wallyworld> bradm: status is somewhat inefficient, especially on large models. there are plans to fix it but right now there's not much that can be done. it is worse if the controller is dealing with a lot of traffic like config change processing etc. it is better in juju 2.2.2 than other versions but still not great
[04:21] <bradm> wallyworld: this is 2.2.2, about 50 machines
[04:21] <bradm> wallyworld: and about 80 lxds on the machines it looks like
[04:21] <wallyworld> it's more the number of apps and units
[04:21] <bradm> its an openstack deploy, so its not tiny
[04:22] <wallyworld> that would do it
[04:23] <wallyworld> sadly there's nothing that can be done until a fix lands in a future release
[04:24] <bradm> it seems like a machine-0 restart helps for a bit, but it gets worse over time
[04:25] <wallyworld> sadly yes
[04:25] <wallyworld> veebers: tests passed but merge failed :-( http://juju-ci.vapour.ws:8080/blue/organizations/jenkins/github-merge-juju/detail/github-merge-juju/97/pipeline
[04:26] <bradm> the other problem is out of all those hundred or so units, one has decided its going to use its ipv6 address for juju, is there any way to fix that?
[04:33] <veebers> wallyworld: looking
[04:41] <veebers> wallyworld: rats, sorry about that, need to re-run it now. But the merge will work this time around
[04:42] <wallyworld> ok
[05:01] <veebers> wallyworld: that's done now
[05:01] <wallyworld> great ty
[05:02] <veebers> wallyworld: nw, sorry for the hassle there
[05:02] <wallyworld> shit happens
[05:44] <wallyworld> babbageclunk: maybe i can swap you reviews? this is proposed against a branch of my fork because i'm waiting on a review of that upstream branch before it can be landed https://github.com/wallyworld/juju/pull/44
[05:44] <babbageclunk> wallyworld: sure!
[05:44] <babbageclunk> wallyworld: also I need to do one for axw
[05:45] <wallyworld> so many reviews - we must be doing work or something
[06:06] <wallyworld> babbageclunk: lgtm but with a request to change the constructor - hope it makes sense
[06:10] <wallyworld> axw: witg bug 1706535, to refresh my memory, it's just a matter of setting the upgrade step against 2.2.3 version i think?
[06:10] <mup> Bug #1706535: Upgrading juju 2.2.1 to 2.2.2 fails on Azure and vSphere <juju:Triaged by axwalk> <https://launchpad.net/bugs/1706535>
[06:11] <axw> wallyworld: 2.2.2, but otherwise yes
[06:11] <wallyworld> axw: 2.2.2 is released, so it has to be 2.2.3
[06:11] <wallyworld> i guess it doesn't matter really
[06:11] <axw> wallyworld: no... because we don't need 2.2.2 to run the step
[06:12] <axw> i.e. if you upgrade 2.2.2 to 2.2.3, it shouldn't run
[06:12] <wallyworld> right ok. but it will be run from 2.1.x to 2.2.3
[06:12] <axw> wallyworld: right, and (crucially) also 2.2.0 and 2.2.1 to 2.2.3
[06:13] <axw> it was those ones for which it was broken
[06:13] <axw> was/is
[06:13] <axw> brb, vbox wants me to restart
[06:19] <wallyworld> axw: not sure if you'll get a chance to look at the 2nd pr today? where the macaroon auth is wired up? i re-targetted to develop since the other branch landed earlier. maybe in between tasks? thanks for reviewing the first one
[06:20] <axw> wallyworld: not sure yet, I'm flailing with the LXC-on-MAAS testing. I'd like to make some progress here
[06:20] <wallyworld> sure np
[09:35] <babbageclunk> wallyworld: Sorry, I got pulled onto kid duty but I'm going through that PR now - hope I'm not doubling up with axw!
[09:36] <axw> babbageclunk: nope
[09:36] <babbageclunk> cool cool
[09:40] <babbageclunk> axw: I'm looking at your PR next if it's still waiting. :)
[09:40] <axw> babbageclunk: thanks, it is - but it can wait until tomorrow. no rush
[09:40] <babbageclunk> ok
[09:40] <wpk> babbageclunk is on review-frenzy ;)
[09:41] <axw> babbageclunk: I'm trying to get a vMAAS set up, that other one seemed to be a waste of time
[09:41] <babbageclunk> axw: yeah, that architecture thing really stymies it.
[09:42] <babbageclunk> axw: I was thinking you could point it at another (cross-compiled) binary to do the hash, but it's a waste of time unless we actually need to do the upgrade on other archs.
[09:43] <axw> babbageclunk: I did do that, but ended up with other issues that looked like ppc64 bugs. didn't want to spend too much time on it unless I can repro in another MAAS
[09:44] <babbageclunk> oh stink
[09:45] <babbageclunk> wpk: my power level's over 9000!
[22:09] <wallyworld> menn0: if i take a review or 2 off your plate tomorrow, could i ask you to review this for me? https://github.com/juju/juju/pull/7711 also, did you need me to look again at that pr from yesterday? I think your were going to tweak some stuff right
[22:09] <wallyworld> hml: free now for a bit
[22:10] <hml> wallyworld: standup ho?
[22:10] <wallyworld> ok
[22:10] <menn0> wallyworld: will take a look
[22:10] <wallyworld> \o/
[22:10] <menn0> wallyworld: I'll have that other PR updated soon. will ping you.
[22:22] <babbageclunk> wallyworld: hey, did you see my responses on https://github.com/juju/juju/pull/7712?
[22:23] <wallyworld> babbageclunk: i did and as soon as i finish talking to heather was going to ping you
[22:23] <babbageclunk> wallyworld: :)
[22:24] <wallyworld> babbageclunk: wanna jump into the standup ho, almost done
[23:04] <menn0> wallyworld: review done
[23:08] <wallyworld> menn0: awesome just finished talking to eric, looking now. eric will be ready to do stuff for you today or tomorrow
[23:15] <wallyworld> menn0: i have another PR stacked on top of the one you reviewed. i'll make the changes in that one and merge them in one go. otherwise i'll get lots of conflicts
[23:15] <wallyworld> thanks again for looking
[23:25] <anastasiamac> wallyworld: since u r talking to everyone today, could i take a ticket and stnad in line too, plz?