[00:31] <wallyworld> babbageclunk: standup?
[00:31] <babbageclunk> wallyworld: oh, sorry
[02:21] <axw> wallyworld: "The remotefirewaller facade is removed and the ingress address watcher moved to the crossmodelrelations facade."  <- that's meant to say firewaller, not crossmodelrelations, right?
[02:22] <wallyworld> axw: yeah, sorry, typo
[02:23] <wallyworld> i'll correct the pr description
[02:42] <axw> wallyworld: can we go over this PR together next week? does it need to land yet? I'm super foggy and struggling to consume it all at once
[02:43] <wallyworld> axw: no worries, i'll just brnach off it for the next one
[02:43] <wallyworld> the core firewaller change is 100 lines or so but there's the boilerplate all around it
[02:44] <axw> wallyworld: thanks. I've left a few minor comments, just struggling to separate refactor from functional change atm, and don't want to let bits slip through the cracks, especially in the firewaller worker
[02:44] <wallyworld> no worries.
[02:45] <wallyworld> luckily the core of the worker itself for non-cmr is unchanged
[02:45] <wallyworld> plenty of time next week
[02:45] <wallyworld> can even pair review etc
[05:12] <babbageclunk> wallyworld: can you take a look at https://github.com/juju/juju/pull/7542 please?
[05:12] <wallyworld> sure
[05:12] <babbageclunk> wallyworld: QAing it now
[05:19] <thumper> ah fark
[05:21] <wallyworld> babbageclunk: a couple of small things
[05:21] <babbageclunk> wallyworld: thanks
[05:23] <thumper> jam: https://github.com/juju/juju/pull/7543
[05:27] <babbageclunk> wallyworld: The other IsBlahError functions in that file all take interface{} - I'm alright to change mine, but I wonder whether there's a reason for it.
[05:27] <wallyworld> hmmm, not sure, let me look
[05:29] <wallyworld> babbageclunk: i think it's just poor code tbh
[05:29] <wallyworld> i can't see a reason for it
[05:31] <babbageclunk> wallyworld: yeah, looking at other places they all take error - I'll change them.
[05:31] <wallyworld> sgtm
[05:42] <jam> thumper: lgtm
[05:43] <thumper> jam: awesome
[05:48] <thumper> jam: your pinger branch looks good to me, mostly comments about comments
[05:48]  * thumper out
[05:57] <babbageclunk> wallyworld: hmm, QA hitting problems - uniter is throwing errors with the new error message - tracking it down now.
[05:58] <wallyworld> righto
[06:43] <wpk> wallyworld: will you be merging 2.2 into develop anytime soon? There are some CMR conflicts there and I don't want to break it
[06:43] <wallyworld> wpk: yes, as soon as final changes land
[06:43] <wpk> wallyworld: kk
[06:44] <wallyworld> don't worry about breaking thinhs, i can deal with it
[07:05] <wpk> is the proces of moving issues from 'fix commited' to 'fix released' automatical?
[08:01] <wallyworld> wpk: manual, but there's a script they can adapt to help
[08:43] <wpk> https://github.com/juju/cmd/pull/52 anyone?
[09:05] <babbageclunk> wallyworld: I think I've cracked it.
[10:04] <babbageclunk> wa..
[10:04] <babbageclunk> doh
[10:04] <babbageclunk> wallyworld: I have not cracked it, I'm afraid. I need to discuss with you about it tomorrow. Sorry!
[10:14] <wallyworld> babbageclunk: what's theissue
[10:16] <babbageclunk> wallyworld: I keep finding more places that assume you can call relation.Unit for any unit of an application, which isn't really right, but I've made it an error.
[10:16] <wallyworld> ah
[10:17] <wallyworld> babbageclunk: so worst case, we just need to tell people to upgrade to 2.2.0 first
[10:17] <wallyworld> and we haven't made it worse since it's broken anyway right now
[10:17] <babbageclunk> It's not right because those RelationUnits are nonsensical - eg a wordpress subordinate logging unit in a mysql-logging relation.
[10:17] <babbageclunk> Rught
[10:17] <babbageclunk> Oops, right
[10:18] <babbageclunk> but my PR is a fair bit away from being releasable
[10:18] <babbageclunk> I mean, the one that isn't landed yet.
[10:26] <wallyworld> babbageclunk: right, but in 2.2.1 everything is fixed right?
[10:26] <wallyworld> so the only issue is people upgrading from 2.1.x
[10:26] <wallyworld> and if they go via 2.2.0 they will be ok
[10:27] <wallyworld> or have i missed something?
[10:38] <babbageclunk> wallyworld: yeah, I think so - the errors I see are coming from things like unit.RelationsInScope, which constructs nonsensical RUs to call InScope on them, in which case it returns false so everything's fine.
[10:40] <wallyworld> babbageclunk: ok. ideally of course we'd be able to go from 2.1.x straight to 2.2.1 but at least we have a work around
[10:40] <wallyworld> we should still get the fix done asap
[10:41] <babbageclunk> wallyworld: So maybe rather than having relation.Unit throw an error (which breaks lots of places that currently create nonsensical RUs), I could add an IsValid() and check that in the uniter API EnterScope.
[10:42] <babbageclunk> wallyworld: make the sensibility check opt-in.
[10:42] <wallyworld> babbageclunk: hmmm, that might work. is that unit API enterscope the the only vector via which the upgrade issue manifests itself?
[10:43] <babbageclunk> wallyworld: I think so - might just be the only place I/we've noticed it.
[10:43] <babbageclunk> wallyworld: I think I'm too zonked to think about it now though.
[10:45] <wallyworld> babbageclunk: no worries
[10:46] <babbageclunk> wallyworld: night!
[10:47] <wallyworld> ttyl
[22:28] <thumper> babbageclunk: is bug 1696509 fixed?
[22:28] <mup> Bug #1696509: status-history-pruner fails under load <adrastea> <performance> <pruning> <statuseshistory> <juju:In Progress by 2-xtian> <https://launchpad.net/bugs/1696509>
[22:28] <babbageclunk> thumper: nope - I haven't had a chance to do the load testing that jam wants yet.
[22:28] <thumper> ok
[22:28] <thumper> I'll push it off
[22:47] <babbageclunk> wallyworld: hey, so do you think I should pursue that IsValid thing I was talking about yesterday?
[22:48] <wallyworld> babbageclunk: we were'nt going to but now it looks like a race needs fiing
[22:48] <wallyworld> maybe see if there's a quick win
[22:49] <babbageclunk> wallyworld: ok - the race is holding up the release anyway?
[22:50] <wallyworld> at this stage, yeah