[00:44] <davechen1y> lucky(~/src/github.com/juju/juju) % gb test
[00:44] <davechen1y> FATAL command "test" failed: failed to resolve import path "github.com/juju/juju/api": cannot find package "golang.org/x/net/http2/hpack" in any of:
[00:44] <davechen1y> ^ who added a dependency on the new http2 stuff ?
[00:48] <thumper> axw: yes
[00:48] <thumper> I believe so
[00:48]  * thumper shrugs
[00:49] <thumper> davechen1y: the shrug was for the http2 stuff
[00:50] <axw> davechen1y: "go list" shows no dependency on http2
[00:50] <thumper> perhaps just a test dep?
[00:51] <davechen1y> could be a test dep
[00:51] <davechen1y> still digging
[00:57] <davechen1y> ok
[00:57] <davechen1y> ignore this
[00:58] <davechen1y> short version: it's a go 1.6 thing
[00:58] <davechen1y> ignore it
[01:05] <thumper> :)
[01:08] <mup> Bug #1510333 opened: worker/uniter: converge on a single approach for updating goal state <tech-debt> <juju-core:Triaged> <https://launchpad.net/bugs/1510333>
[01:14] <mup> Bug #1510333 changed: worker/uniter: converge on a single approach for updating goal state <tech-debt> <juju-core:Triaged> <https://launchpad.net/bugs/1510333>
[01:17] <mup> Bug #1510333 opened: worker/uniter: converge on a single approach for updating goal state <tech-debt> <juju-core:Triaged> <https://launchpad.net/bugs/1510333>
[01:29] <davechen1y> http://reviews.vapour.ws/r/3000/
[01:29] <davechen1y> anyone ?
[01:31] <thumper> davechen1y: this is from where?
[01:31] <thumper> was it juju/juju/ssh?
[01:31] <thumper> I was going to suggest we just have juju/ssh rather than juju/utils/ssh
[01:31] <thumper> but then I saw the other deps on the utils package
[01:31] <thumper> and thought "meh"
[01:33] <davechen1y> thumper: the only thing worse than a utils packge is a utils repo
[01:33] <davechen1y> thumper: and the only thing worse than a utils repo is having several
[01:33] <davechen1y> one evil at a time
[01:33] <thumper> :)
[01:33] <thumper> k
[01:36] <thumper> axw: if you are happy with https://github.com/juju/retry/pull/1 I'll merge it :)
[01:36] <axw> thumper: sorry, I keep getting distracted
[01:36] <thumper> axw: I was hoping fwereade would have looked, but I'm prepared to land it now
[01:36] <axw> thumper: lemme take another quick look
[01:36] <thumper> cheers
[01:39] <sinzui> thumper: I have added github.com/juju/retry. We wont know it works until there is a branch to merge. Since there isn't a branch there now, I don't know which packages deps it imports.
[01:39] <thumper> sinzui: https://github.com/juju/retry/pull/1
[01:40] <sinzui> thank youn thumper . I will quickly add the deps I see
[01:40] <thumper> cheers
[01:42] <axw> thumper: reviewed
[01:43] <thumper> axw: ta
[01:51] <thumper> sinzui: I'll try merging now
[01:51] <thumper> sinzui: is it ready?
[01:51] <sinzui> thumper: yep, I am watching to procs now
[01:52] <thumper> sinzui: who handles which repos go to reviewboard?
[01:54] <sinzui> thumper: I don't know the config. I am sure ericsnow knows.
[01:55] <thumper> \o/ landed
[01:55] <sinzui> thumper: I see real go test was really run too
[01:55] <thumper> as opposed to what?
[01:56] <sinzui> thumper: the merge job just sending a success without actually merging the branch and running the tests
[01:56] <thumper> :)
[01:57] <thumper> I must admit, I do find it pleasing to see more tests than code :)
[01:57] <sinzui> :))
[02:53] <davechen1y>                         logger.Debugf("keeping unrecognised existing ssh key %q: %v", key, err)
[02:53] <davechen1y> ERRORS LOGGED AT DEBUGGG!
[02:54] <anastasiamac> "some error are more equal than others"? :P
[02:55] <davechen1y> "some errors are more like guidelines"
[02:57] <anastasiamac> :P
[03:01] <davechen1y> thumper: FFS
[03:01] <thumper> davechen1y: hey
[03:01] <davechen1y> tried to apply your suggestions to use the errors package, all the fucking tests expect an exact string
[03:01] <davechen1y> tempted to tell you to stick your recommendation
[03:01] <thumper> davechen1y: that's fine for now
[03:01] <davechen1y> cannot even upgrade them to errors.Annotate
[03:01] <thumper> we can "fix" it later
[03:01] <davechen1y> heh, fix
[03:16] <natefinch> Man, can't tell you how much I hate all our gc.ErrorMatches tests.  What a nightmare.
[03:17] <davechen1y> i fixed it, with a hammer
[03:25] <davechen1y> axw: re ExecuteCommandOnMachine
[03:25] <davechen1y> what if it went to juju/utils/exec
[03:25] <axw> davechen1y: SGTM
[03:25] <axw> davechen1y: oh hum, but htat means utils/exec depends on ssh?
[03:27] <axw> davechen1y: if it's a PITA, don't worry. I'd just like for hte ssh package to be a bit leaner
[03:30] <davechen1y> i think i can fix that
[03:30] <davechen1y> type Client interface {
[03:30] <davechen1y> type Client interface {
[03:30] <davechen1y>  // Client is an interface for SSH clients to implement
[03:30] <davechen1y> type Client interface {
[03:31] <davechen1y> ^ honestly, it should just say "client is an interface"
[03:31] <davechen1y> for all the good that comment does :)
[03:32] <davechen1y> axw, hmm
[03:32] <davechen1y> this is quite tricky
[03:32] <davechen1y> execonmachien requires something that mathces Command(...) *ssh.Cmd
[03:32] <davechen1y> dunno if it can be teased apart easily
[03:33] <axw> davechen1y: yeah, just drop it. maybe it can be refactored later, not worth blocking the move over
[03:34] <davechen1y> mkay
[03:39] <davechen1y> i'm going to loose my shit this week http://juju-ci.vapour.ws:8080/job/github-merge-juju-utils/47/console
[03:39] <davechen1y> ^ anyone want to guess why this broke ?
[03:45] <axw> davechen1y: because it's using go 1.2 I'm guessing
[03:46] <axw> davechen1y: (and we're trying to compile it with that when we should just be downloading it and then checking out the older rev)
[03:54] <davechen1y> axw: i think there might be some sort of conditional compilation going on there
[03:54] <davechen1y> we pin the revision we want in juju/juju
[03:54] <davechen1y> but trying to compile juju/utils, it floats freely
[03:55] <axw> davechen1y: ah hm, cos we don't have a dependencies.tsv there ? :/
[03:55] <davechen1y> i dunno if adding one automatically help
[03:55] <axw> bit of an oversight
[03:55]  * axw shrugs
[03:56] <davechen1y> here's what I propose
[03:56] <davechen1y> 1. i go an have some lunch
[03:56] <davechen1y> 2. it magically fixes itselp
[03:56] <davechen1y> sgtm ?
[03:56] <axw> :)
[03:56] <axw> sounds good. I'll do the same
[03:56] <natefinch> 1. switch to gb
[03:56] <natefinch> 2. ???
[03:56] <natefinch> 3. profit
[09:32] <dimitern> frobware, hey, I'll be a couple of minutes late for our 1:!
[09:32] <dimitern> 1:1
[09:35] <frobware> dimitern, ok
[10:02] <dimitern> jam, fwereade, standup?
=== Makyo is now known as Guest72537
[10:36] <axw> fwereade: my latest rev on http://reviews.vapour.ws/r/2935/ is quite substantial, I'd appreciate a look at the last commit when you can
[10:37] <axw> fwereade: there's an unrelated fix mised in there, but it's also up for review as http://reviews.vapour.ws/r/3002/
[10:37] <axw> s/mised/mixed/
[10:47] <fwereade> axw, thanks
[11:13] <fwereade> axw, both reviewed
[12:05] <fwereade> https://github.com/juju/juju/wiki/howto:-implement-effective-config-structs
[12:05] <fwereade> ...if anyone feels like some light reading or arguing
[12:56] <wallyworld> fwereade: did you have time for a chat?
[12:58] <fwereade> wallyworld, just about to have lunch -- what time is it for you?
[12:58] <wallyworld> 23:00, i don't mind waiting
[12:59] <wallyworld> just ping when you're free
[13:07] <mup> Bug #1510533 opened: destroy-environment panics <juju-core:New> <https://launchpad.net/bugs/1510533>
[13:10] <mup> Bug #1510533 changed: destroy-environment panics <juju-core:New> <https://launchpad.net/bugs/1510533>
[13:13] <mup> Bug #1510533 opened: destroy-environment panics <juju-core:New> <https://launchpad.net/bugs/1510533>
[14:02] <katco> ericsnow: wwitzel3: natefinch: please do the standup without me. cats shattered lots of glass on our kitchen floor and my dog ate some
[14:02] <perrito666> katco: ate some cat or glass?
[14:04] <wwitzel3> katco: ok
[14:05] <frobware> dimitern, ping
=== marlinc_ is now known as marlinc
[14:43] <abentley> sinzui: chat?
[14:44] <sinzui> abentley: sure
[15:26] <mup> Bug #1403689 changed: Server should handle tools of unknown or unsupported series <upgrade-juju> <upload-tools> <juju-core:Fix Released by wallyworld> <juju-core 1.24:Triaged> <juju-core 1.25:Fix Released by wallyworld> <https://launchpad.net/bugs/1403689>
=== Guest72537 is now known as Makyo
[16:32] <mup> Bug #1510619 opened: Wily: add machine fails using kvm and lxcbr0 <add-machine> <cloud-installer> <kvm> <network> <juju-core:New> <https://launchpad.net/bugs/1510619>
=== akhavr1 is now known as akhavr
[17:40] <voidspace> mgz_: ping
[17:40] <mgz_> voidspace: wotcha
[17:40] <voidspace> mgz_: hey
[17:41] <voidspace> mgz_: so we have plans to get rid of the need for ignore-addresses on master
[17:41] <mgz_> ace
[17:41] <voidspace> mgz_: but we're going to land the fix and test it with those needing it *before* we actually remove ignore-addresses
[17:41] <voidspace> (I work on the proposed fix next)
[17:41] <voidspace> mgz_: so we'll land the change to ignore-addresses on containers and master first
[17:42] <voidspace> mgz_: because currently containers are broken when you use ignore-machine-addresses
[17:42] <voidspace> mgz_: and next a question...
[17:42] <mgz_> voidspace: sounds fair
[17:42] <voidspace> mgz_: when I attempt to do an upgrade from 120/1.22 to 1.24 but using --upload-tools
[17:42] <voidspace> mgz_: it fails with "unknown series wily"
[17:43] <voidspace> mgz_: dimitern thought you might have encountered this, or similar, when testing upgrades with QA and know a workaround
[17:45] <mgz_> voidspace: yeah, we've had a lot of that in testing,
[17:45] <voidspace> mgz_: did you solve it?
[17:45] <mgz_> not sure if I know a specific workaround when using --upload-tools for upgrade though
[17:45] <voidspace> :-/
[17:46] <voidspace> I'll ask aaron
[17:46] <voidspace> abentley: ping
[17:46] <mgz_> bug  1403689 is part of it
[17:46] <voidspace> bug 1403689
[17:46] <mup> Bug #1403689: Server should handle tools of unknown or unsupported series <upgrade-juju> <upload-tools> <juju-core:Fix Released by wallyworld> <juju-core 1.24:Triaged> <juju-core 1.25:Fix Released by wallyworld> <https://launchpad.net/bugs/1403689>
[17:46] <voidspace> mgz_: ah, I wondered if there was a bug for it
[17:46] <voidspace> mgz_: thanks
[17:47] <voidspace> unassigned for 1.24 currently
[17:47] <mgz_> what CI does is actually use streams, which we then control
[17:47] <voidspace> right
[17:47] <voidspace> I thought as much, and we can't suggest that to IS I don't think
[17:47] <voidspace> we want them to test the new binaries
[17:47] <mgz_> but yeah, bug suggests --upload-tools to 1.25 would work?
[17:48] <voidspace> 1.25 isn't released yet
[17:48] <mgz_> upgrades with --upload-tools are pretty dodgy anyway
[17:48] <voidspace> needed when you have new binaries for a user to test
[17:49] <sinzui> voidspace: your experience mirror's abentley's frustration over the past 48 hours
[17:50] <voidspace> sinzui: heh
[17:50] <voidspace> ouch
[17:50] <sinzui> voidspace: the stupid client makes assumptions about its host to decide what to upload to the series and since 1.20.x doesn't check distro-info-data correctly, it will die because of vivid or wily
[17:51] <voidspace> right :-/
[17:51] <voidspace> sinzui: it makes testing upgrades / fixes for upgrade bugs impossible
[17:51] <voidspace> hopefully the fix is correct, so we never have the issue again
[17:53] <sinzui> voidspace: abentley  might want to bu you beer
[17:53] <voidspace> sinzui: hah
[17:54] <voidspace> sinzui: no, I meant the fix that wallyworld has allegedly committed to 1.25 and master
[17:54] <voidspace> not that I've done it
[17:54] <voidspace> he hasn't committed it to 1.24 though, which is where we need it
[17:54] <sinzui> voidspace: CI tests 1.20 upgrades to the candidate we test. abentley is updating the test to use upload-tools as needed. so your change is exactly what we are looking to test
[17:54] <voidspace> right
[17:54] <voidspace> so we need 1.20 and onwards fixed really
[17:54] <voidspace> might be worth abentley pinging wallyworld about it
[17:55] <sinzui> voidspace: yeah, we can do a 1.20 release if we need to. We just need to make several jobs non-voting before hand to check to deal with unsupported features
[17:56] <voidspace> sinzui: although a 1.20 to 1.22 upgrade works I believe, 1.24 is the problematic version
[17:56] <voidspace> sinzui: so just fixing 1.22 on would suffice
[17:56] <voidspace> which is one less release at least
[17:57] <sinzui> voidspace: we can test/release 1.22 at anytime. we have special PPA setup for its distrubution too
[17:57] <voidspace> cool
[17:58] <voidspace> I've added a comment to the bug
[18:05] <mup> Bug #1510651 opened: Agents are "lost" after terminating a state server in an HA env <juju-core:New> <https://launchpad.net/bugs/1510651>
[18:29] <voidspace> frobware: our proposed fix for getting rid of ignore-machine-addresses can't work
[18:29] <voidspace> frobware: it's effectively what we're already doing
[18:29] <voidspace> frobware: we might need to prefer a provider fallback address over an exact match machine address
[18:29] <voidspace> frobware: I'll discuss it with dimiter tomorrow
[18:30] <voidspace> frobware: meanwhile, the fix for containers has landed on 1.25 and is landing on master
[18:30] <voidspace> frobware: and I've tested on both 1.25 and master
[18:32] <abentley> voidspace: pong
[18:36] <voidspace> abentley: unping
[18:36] <voidspace> abentley: it's about the wily issue and upgrades
[18:36] <voidspace> abentley: see the above conversation if you want... :-)
[18:36] <voidspace> abentley: I'll try and persuade wallyworld that we need the fix for 1.22 & 1.24
[18:39] <abentley> voidspace: I'm not sure whether we need 1.22.  1.24 would be nice, given that we may need to do a 1.24.8 at some point.
[18:40] <voidspace> abentley: well, if you want to test an upgrade *from* 1.22 (as I do as it happens and I thought you needed to) then we need it in 1.22
[18:41] <abentley> voidspace: Mwuh?  You mean he fixed the server instead of the client?
[18:42] <voidspace> abentley: the error message is from the api server
[18:42] <voidspace> not a client issue as far as I know
[18:43] <voidspace> abentley: see bug 1403689
[18:43] <mup> Bug #1403689: Server should handle tools of unknown or unsupported series <upgrade-juju> <upload-tools> <juju-core:Fix Released by wallyworld> <juju-core 1.24:Triaged> <juju-core 1.25:Fix Released by wallyworld> <https://launchpad.net/bugs/1403689>
[18:43] <voidspace> abentley: it's the api server that rejects the upload
[18:46] <abentley> voidspace: The client should not try to upload tools that it knows will be rejected by the server.
[18:47] <natefinch> katco: is your dog ok?
[18:50] <voidspace> abentley: not sure how it was fixed
[18:53] <mup> Bug #1510619 changed: Wily: add machine fails using kvm and lxcbr0 <add-machine> <cloud-installer> <kvm> <network> <juju-core:Invalid> <lxc (Ubuntu):New> <https://launchpad.net/bugs/1510619>
[18:59] <mup> Bug #1510619 opened: Wily: add machine fails using kvm and lxcbr0 <add-machine> <cloud-installer> <kvm> <network> <juju-core:Invalid> <lxc (Ubuntu):New> <https://launchpad.net/bugs/1510619>
[19:05] <mup> Bug #1510619 changed: Wily: add machine fails using kvm and lxcbr0 <add-machine> <cloud-installer> <kvm> <network> <juju-core:Invalid> <lxc (Ubuntu):New> <https://launchpad.net/bugs/1510619>
=== natefinch is now known as natefinch-afk
[19:44] <mup> Bug #1510675 opened: vSphere provider leaves behind virtual machines upon bootstrap failure <bootstrap> <vsphere> <juju-core:New> <https://launchpad.net/bugs/1510675>
[19:47] <mup> Bug #1510675 changed: vSphere provider leaves behind virtual machines upon bootstrap failure <bootstrap> <vsphere> <juju-core:New> <https://launchpad.net/bugs/1510675>
[19:50] <mup> Bug #1510675 opened: vSphere provider leaves behind virtual machines upon bootstrap failure <bootstrap> <vsphere> <juju-core:New> <https://launchpad.net/bugs/1510675>
[20:17] <mup> Bug #1510688 opened: sync-tools tries to upload agents that are not permitted by the state server <sync-tools> <juju-core:Triaged> <https://launchpad.net/bugs/1510688>
[20:17] <mup> Bug #1510689 opened: juju upgrade --upload-tools tries to upload tools agents that are not permitted by the state server <upgrade-juju> <upload-tools> <juju-core:Triaged> <https://launchpad.net/bugs/1510689>
[20:20] <mup> Bug #1510688 changed: sync-tools tries to upload agents that are not permitted by the state server <sync-tools> <juju-core:Triaged> <https://launchpad.net/bugs/1510688>
[20:20] <mup> Bug #1510689 changed: juju upgrade --upload-tools tries to upload tools agents that are not permitted by the state server <upgrade-juju> <upload-tools> <juju-core:Triaged> <https://launchpad.net/bugs/1510689>
[20:26] <mup> Bug #1510688 opened: sync-tools tries to upload agents that are not permitted by the state server <sync-tools> <juju-core:Triaged> <https://launchpad.net/bugs/1510688>
[20:26] <mup> Bug #1510689 opened: juju upgrade --upload-tools tries to upload tools agents that are not permitted by the state server <upgrade-juju> <upload-tools> <juju-core:Triaged> <https://launchpad.net/bugs/1510689>
[21:59] <wallyworld> cmars: running a bit late, can i ping you
[21:59] <cmars> wallyworld, sure, np
[22:07] <alexisb> cmars, that was my fault, that wallyworld was late
[22:08] <cmars> alexisb, heh no worries
[22:08] <wallyworld> cmars: it's always alexisb fault
[22:08] <wallyworld> there now
[22:36] <wallyworld> axw: anastasiamac: perrito666: i have to make an emergency run to the school to drop off some stuff, so will miss standup; i'll talk to you guys when i get back to catch up
[22:41] <anastasiamac> wallyworld: k
[22:49] <perrito666> wallyworld: k