[00:05] <babbageclunk> wallyworld: awesome, thanks
[01:36] <thumper> I have found a bunch of failures where people were creating state pools unnecessarily in tests
[01:36] <thumper> I'm going to land those fixes independently
[01:46] <thumper> babbageclunk, axw or wallyworld: https://github.com/juju/juju/pull/8189
[01:46] <wallyworld> looking
[01:47] <thumper> wallyworld: I'm trying to reduce the surface of the big pull request for the watchers
[01:47]  * wallyworld nods
[01:48] <babbageclunk> thumper: lgtm
[01:48] <babbageclunk> doh, sorry
[01:49] <thumper> all good
[01:49]  * thumper afk briefly to collect daughter
[02:11] <thumper> axw: any thoughts on https://bugs.launchpad.net/juju/+bug/1736582 ?
[02:11] <mup> Bug #1736582: Cannot bootstrap manual provider with Juju 2.2.6 <juju:New> <https://launchpad.net/bugs/1736582>
[02:53] <thumper> babbageclunk: another https://github.com/juju/juju/pull/8190
[02:58] <thumper> or wallyworld: ^^
[03:07] <axw> thumper: sorry missed your q. from casey's latest comment, sounds like we should perform a check to see if the mongo port is firewalled
[03:09]  * thumper nods
[03:12] <wallyworld> thumper: looking
[03:12] <thumper> wallyworld: thanks, this one is boring too
[03:12] <wallyworld> axw: here's that pr we talked about https://github.com/juju/juju/pull/8191
[03:13] <axw> wallyworld: ok, will take a look later after lunch
[03:13] <wallyworld> yup, no rush
[03:14] <wallyworld> thumper: lgtm, interfaces good
[03:14] <thumper> wallyworld: ta
[04:40] <axw> wallyworld: how come you only pick out one port?
[04:41] <wallyworld> let me look
[04:41] <axw> wallyworld: in EnsureService
[04:43] <axw> wallyworld: I think it makes sense to have ingress just point at the first container port in the spec (we can have that as a convention), but the service can still expose all ports?
[04:43] <wallyworld> axw: i guess no good reason, just the current assumption that we only support one container per pod
[04:43] <wallyworld> yeah
[04:43] <wallyworld> agreed, that is easily fixed
[04:43] <axw> cool
[04:44] <axw> wallyworld: though I'm not sure if there's a way to determine which one is "first" once you fetch the service definition. is order preserved?
[04:44] <axw> wallyworld: i.e. in ExposeService, is the order of ports there the same as what was specified in EnsureService
[05:18] <wallyworld> axw: thanks for review. i think the order is preserved. at least frm what i have seen
[05:41] <axw> wallyworld: ok cool
[05:42] <axw> wallyworld: I think it would be best to do the conversion to int32 internally
[05:42] <wallyworld> yup, done
[06:23] <wallyworld> axw: awesome, go 1.10 beta1 changes go fmt behaviour :-/
[06:25] <axw> wallyworld: ah well, we'd better all upgrade then
[06:25] <wallyworld> and the builders :-)
[06:25] <axw> oh right, we check that don't we
[08:08] <wallyworld> axw: here's the caas firewaller all completed https://github.com/juju/juju/pull/8193
[08:09] <axw> wallyworld: I think that's going to be a monday job, my brain is dead
[08:09] <axw> I'll make a start on it anyway
[08:10] <wallyworld> no worries, no rush
[08:11] <wallyworld> i'm just happy to get it all working
[08:11] <wallyworld> i'm off to have dinner, have a good weekebd
[08:11] <axw> wallyworld: yeah, that's a solid effort - wasn't expecting to see it so soon
[08:11] <axw> wallyworld: you too
[10:35] <wallyworld> axw: thanks for the fix. LGTM that should have been caught in CI. clearly there's something missing from the upgrade tests
[10:35] <axw> wallyworld: ta
[10:35] <axw> yep
[10:35] <wallyworld> jam: did you have any thoughts on that network info issue