[00:49] <veebers> wallyworld: migrating a model to a controller of a newer version is supported right?
[00:50] <veebers> thumper: can you share the 2.3.6 upgrade bug link please?
[00:51] <wallyworld> veebers: yep
[00:52] <anastasiamac> veebers: 2.3.6 upgrade bug - https://bugs.launchpad.net/bugs/1765722
[00:52] <mup> Bug #1765722: upgrade to 2.3.6 failed: the dotted field is not valid for storage <docteam> <openstack> <sts> <juju:In Progress by jameinel> <juju 2.3:In Progress by jameinel> <https://launchpad.net/bugs/1765722>
[00:52] <veebers> wallyworld: ok, we might have a bug then :-\ https://paste.ubuntu.com/p/pmt4cfYd64/
[00:52] <veebers> awesome, cheers an
[00:52] <veebers> anastasiamac ^ :-)
[00:52] <anastasiamac> haha, yes let's NOT abbreviate!! :D
[00:52] <wallyworld> veebers: oh, i forgot, we don't yet support migrating models with offers
[00:53] <veebers> wallyworld: ah, hah ok. No bug then, just unsupported :-|
=== menn0_ is now known as menn0
[00:53] <wallyworld> veebers: yeah, cause if you migrate a model with offers, things connected to them will break
[00:54] <veebers> wallyworld: ack, makes sense. Have I misunderstood the macaroon testing plan from Friday, or does this throw a spanner in the works?
[00:55] <wallyworld> veebers: no, i just forgot about the migration limitation
[00:56] <veebers> wallyworld: ack, I'll shelve for now and get this initial tomb PR up.
[00:56] <wallyworld> yup
[00:58] <veebers> huh, this is concerning: juju destroy-controller mbakev2-offer . . . Continue? (y/N):ERROR controller destruction aborted: read /dev/stdin: resource temporarily unavailable
[01:09] <vino_> wallyworld: have a min
[01:09] <wallyworld> vino_: yeah, just finishing testing, it fails with this error
[01:09] <wallyworld> 2018-04-23 01:08:31 ERROR juju.service.systemd service.go:134 failed to read conf from systemd for application "juju-db": get conf failed (cat /lib/systemd/juju-init/juju-db/juju-db.service): error executing "/bin/systemctl": cat: /lib/systemd/juju-init/juju-db/juju-db.service: No such file or directory;
[01:09] <wallyworld> 2018-04-23 01:08:31 ERROR juju.worker.dependency engine.go:551 "state" manifold worker returned unexpected error: failed to read conf from systemd for application "juju-db": get conf failed (cat /lib/systemd/juju-init/juju-db/juju-db.service): error executing "/bin/systemctl": cat: /lib/systemd/juju-init/juju-db/juju-db.service: No such file or directory;
[01:09] <wallyworld> let's jump into standup ho
[01:09] <vino_> yes please.
[01:12] <kelvinliu> wallyworld, I think i found something different. Do u have a few minutes on hangout?
[01:13] <thumper> wallyworld: can you file a bug about migrating models with cmr?
[01:13] <wallyworld> kelvinliu: give me a few minutes, talking to vinu
[01:13] <wallyworld> thumper: ok. it's something we need to figure out how to handle, won't be trivial
[01:14] <thumper> yeah...
[01:14]  * thumper needs food badly
[01:14] <kelvinliu> wallyworld, all good, let me know when u r free thx
[01:43] <veebers> babbageclunk: I'm seeing 'go vet' 'lock copy' errors on develop. It doesn't seem like the file has changed recently. You re-enabled that check and fixed the issues thought right?
[01:44] <babbageclunk> veebers: yup yup
[01:44] <babbageclunk> Hang on, looking
[01:44] <veebers> babbageclunk: am I going crazy? :-P  Seeing the issue in mongo/mongometrics/mgostatsmetrics.go
[01:45] <babbageclunk> hmm, I'm getting gofmt probs before the go vet.
[01:46] <veebers> babbageclunk: gofmt probs are recent, they are in progress to revert :-)
[01:46] <veebers> I've since fixed the check and merge jobs so they actually fail on go fmt/vet errors
[01:49] <babbageclunk> veebers: running verify.bash doesn't fail once I comment the gofmt check out. It passes the go vet check. Maybe that code isn't actually being checked - is the package used?
[01:51] <babbageclunk> If I run `go tool vet -all .` in mongo/mongometrics I don't get any errors.
[01:52] <veebers> I wonder if I'm doing something odd
[01:52] <veebers> babbageclunk: cheers I'll double check I'm not doing something screwy on my end.
[01:53] <babbageclunk> paste the error you're seeing maybe?
[01:54] <veebers> babbageclunk: https://paste.ubuntu.com/p/63SRG6JCH9/
[01:57] <babbageclunk> veebers: have you updated mgo?
[01:58] <babbageclunk> veebers: in my code, line 38 is copying a struct that only has int fields, but I can see that the package it's in has a global mutex. If the mutex has been moved into the struct that would do it.
[01:59] <thumper> it's always the case where you think you have a nice simple bug, then it all goes to pot
[01:59] <anastasiamac> thumper: the bug u picked did not look simple...
[01:59] <thumper> i thought it might have been
[02:00] <anastasiamac> anastasiamac's definition of simple bug: https://bugs.launchpad.net/juju/+bug/1765688
[02:00] <mup> Bug #1765688: create-storage-pool unnecessarily requires key=value pairs <juju:New> <https://launchpad.net/bugs/1765688>
[02:00] <veebers> babbageclunk: what rev of gopkg.in/mgo.v2 do you currently have?
[02:00] <thumper> https://github.com/juju/juju/blob/develop/apiserver/facades/client/application/deploy.go#L171
[02:00] <anastasiamac> hahaha, awesome comment :)
[02:01] <babbageclunk> f2b6f6c918c452ad107eec89615f074e3bd80e33
[02:01] <babbageclunk> veebers: ^
[02:01] <veebers> babbageclunk: huh, yeah samsies, is yours patched via our make add-patches?
[02:02] <babbageclunk> no, it won't be - I generally don't run make release
[02:02] <babbageclunk> Oh, this is in the build?
[02:02] <veebers> babbageclunk: I just reverted mine (as I had add-patched it), but still failure. I'm confused. I need to look harder I think
[02:02] <babbageclunk> That would explain it
[02:03] <babbageclunk> Yeah, it's the stats race patch
[02:06] <wallyworld> kelvinliu: free now for a chat if you need
[02:06]  * thumper quitely extracts himself from this bug
[02:06] <babbageclunk> veebers: that means the vet error is highlighting a real problem.
[02:06] <veebers> babbageclunk: sigh, yeah passes for me on a second run.
[02:07] <veebers> babbageclunk: right, introduced by the patch
[02:13] <babbageclunk> veebers: I don't think it's too hard to fix - the code needs to pass around *Statses rather than Statses.
[02:14] <babbageclunk> veebers: (and the patch needs to make GetStats return *Stats too)
[02:15] <veebers> babbageclunk: ack, I'll take a go at it shortly. Just trying to determine now why the merge/check jobs failed to pick itup
[02:30] <veebers> babbageclunk: I'm a little confused now :-\ I cleared out those dirs (mgo.v2 etc.) did godeps, applied patches and verify no longer complains. The issue is still there though right? passing around mgo.Stats
[02:31] <babbageclunk> veebers: yeah that's weird - if you apply that one patch, you should be able to see the lock in the Stats struct
[02:52] <wallyworld> thumper: got a minute to join a team standup hangout?
[02:52] <thumper> wallyworld: ack
[02:58] <thumper> veebers: can you pull 2.3.6 out of streams?
[02:59] <veebers> thumper: technically yes? I'm sure it's possible I'm not sure how to do it off the top of my head, would have to have a look
[02:59] <thumper> veebers: please look
[02:59] <veebers> thumper: ack, will do
[03:03] <babbageclunk> thumper: have a moment?
[03:03] <thumper> babbageclunk: kinda
[03:04] <thumper> babbageclunk: 1:1?
[03:04] <babbageclunk> thumper: sure thanks
[05:01] <thumper> anyone? https://github.com/juju/bundlechanges/pull/35
[05:02] <babbageclunk> thumper: oh go on then
[05:10] <thumper> babbageclunk: are you still looking at it?
[05:10] <babbageclunk> thumper: yeah, sorry - also having a discussion on irc.
[05:11] <thumper> :)
[05:12] <babbageclunk> Sorry done now
[05:12] <thumper> ta
[05:45] <veebers> babbageclunk: (having circled back) so odd, *now* I see the govet issues when I run verify script. Any way, would one edit the current patch? maybe spin a new one using the existing as a basis?
[05:46] <babbageclunk> veebers: Not sure - I think editing the existing one is better (less confusing to debug later).
[05:47] <babbageclunk> veebers: it's a bit iffy though - this changes the public interface of the package. But the previous change *kind of* already did that by adding a mutex to something that's returned by value.
[05:48] <veebers> babbageclunk: ack, I'll attack that tomorrow. For now I'll try run the tests I was hoping to run this morning so I can proposed a PR ^_^
[05:48] <veebers> thanks for your help earlier, was scratching my head a bit
[05:48] <babbageclunk> no worries - that's pretty mysterious.
[05:49] <babbageclunk> Hey, we should probably migrate to the fork of mgo! ;)
[05:51] <veebers> ^_^
=== frankban|afk is now known as frankban
[07:16] <vino_> hey WallyWorld : i have few questions
[07:47] <vino_> wallyworld : for the first PR i can verify by myself juju-updateseries.
[07:47] <vino_> Just keeping u posted.
[07:47] <vino_> i need help only for the second PR.
[08:13]  * vino_ away for tea
[08:26]  * vino_ back
[08:36] <wallyworld> vino_: hey
[08:36] <vino_> hey
[08:37] <vino_> i tried reaching u
[08:37] <wallyworld> did you have a PR you wanted me to look at?
[08:37] <vino_> now i am in middle of something...
[08:37] <vino_> i am abt to.
[08:37] <wallyworld> sure, np
[08:37] <vino_> but just one question.
[08:37] <vino_> i cudnt move the agentinfo.c from service to agent folder due to circular package issue..
[08:38] <vino_> more code added and service needs that file and agent needs service.
[08:38] <vino_> so i am moving it to core/agents
[08:38] <vino_> is that fine for u
[08:38] <vino_> ?
[08:38] <wallyworld> ah right, sure, that happens sometimes. i think it may be ok, will be able to take a closer look when i see the PR
[08:38] <vino_> yes.
[08:39] <vino_> and the question i have is.
[08:39] <vino_> inside agents folder..
[08:39] <vino_> i do notice tools dir..
[08:39] <vino_> which i feel can be moved to core/agents/tools
[08:39] <vino_> which is more generic.
[08:39] <vino_> i want to do it and send it the PR.
[08:40] <wallyworld> let me take a quick look
[08:40] <vino_> can u take a look at that folder and let me know ur opinion.
[08:40] <vino_> agents/tools/*
[08:42] <wallyworld> vino_: the packages have slightly different focus. let's leave for now. we can discuss more tomorrow
[08:42] <vino_> oh ok.
[08:42] <vino_> i was doing few chnages
[08:42] <vino_> i will revert.
[08:42] <vino_> so yes.
[08:43] <vino_> we can look at it tomorrow.
[08:43] <vino_> i will send it for review.
[08:43] <wallyworld> the PR will be simpler with just the upgrade bits
[08:43] <vino_> yup.
=== freyes__ is now known as freyes
[23:36] <veebers> wallyworld: re: cmr and migrating, I cannot destroy my controller now (after the failed attempt to migrated a shared model), error is: https://paste.ubuntu.com/p/4PjdrN4BDJ/