[00:28] <davecheney> seriously, the replica set tests never pass on my machine
[00:28] <davecheney> they pass in CI
[00:28] <davecheney> so my care factor is < 1.0
[00:33] <thumper> all tests just passed locally for me
[00:35] <davecheney> yet here we are
[00:37] <davecheney> thumper: i just found a rouge mongodb binary
[00:37] <davecheney> beleted, lets see if that improves things
[00:38] <thumper> :)
[00:38]  * thumper takes the kids and dog out for a walk in the sun
[00:41] <rick_h__> thumper: let me know if you've got a few min please
[01:21] <sebas5384> someone already talked about bundle upgrade?
[01:22] <rick_h__> sebas5384: ?
[01:22] <sebas5384> hey rick_h__!
[01:22] <sebas5384> for example, lets say I deployed a bundle topology into production
[01:23] <sebas5384> and then, i change something, like a new charm related to an existing deployed charm
[01:23] <sebas5384> this is already taking care? by juju ?
[01:23] <rick_h__> sebas5384: yea, to do that we have to update Juju to track and know about services/services that were part of the bundle.
[01:24] <rick_h__> bundles are kind of only an outside idea atm, we're working on pushing the bundle idea deeper into Juju
[01:24] <sebas5384> great to hear that rick_h__
[01:24] <sebas5384> because we are using juju with a CD tool
[01:25] <rick_h__> CD tool?
[01:25] <sebas5384> continues delivery tool
[01:25] <sebas5384> striderCD
[01:25] <rick_h__> ah k
[01:25] <sebas5384> is a CD made in nodejs
[01:25] <rick_h__> ah cool, we almost used that for CI
[01:25] <sebas5384> ooh nice
[01:25] <rick_h__> I had some chats with them about our desired github integration/etc
[01:25] <sebas5384> hmm
[01:26] <sebas5384> yeah they integrate very well with github
[01:26] <sebas5384> but I understand the tool is in early stages
[01:26] <rick_h__> yea
[01:26] <sebas5384> so the thing is we are planning to do a juju plugin
[01:27] <sebas5384> for the stridercd
[01:27] <sebas5384> so you can relate a bundle to a branch
[01:27] <sebas5384> and then to an environment
[01:28] <sebas5384> so if I add a charm into the bundle yaml file, it should update my environments
[01:29] <rick_h__> right, to do that properly, you need to track a lot of info that the env doesn't currently have
[01:29] <rick_h__> we're working on making bundles support revisions and allowing them to live in the charm store right now
[01:29] <sebas5384> yeah I thought about that
[01:29] <rick_h__> it's the first baby step into that data/tracking you're looking for
[01:29] <sebas5384> its related to the revision API ?
[01:29] <sebas5384> nice!
[01:29] <rick_h__> well the idea is that you want to be able to tell a diff from what's changed.
[01:30] <sebas5384> thats right
[01:30] <rick_h__> and the store will be able to at least provide you rev 3 bundle and rev 4. Though really you probably just want to know what's different from your current env and the updated bundle
[01:30] <rick_h__> there's a bunch of that stuff to work out
[01:31] <sebas5384> of course
[01:32] <sebas5384> and I saw the new GUI, with the commit changes
[01:32] <rick_h__> commit changes?
[01:33] <sebas5384> hmmm i don't know how to called it
[01:33] <rick_h__> ah ok. Well if you've got any gui questions I'm your guy on that end
[01:33] <sebas5384> but I sow that we are going to have an stage state, so then you have to "commit" those changes
[01:34] <sebas5384> yeah I know!!! hehe
[01:34] <rick_h__> ah, yes. That's what we call the 'deployer bar' and it is part of the machine view work
[01:34] <rick_h__> so you can run colocation through the GUI
[01:34] <sebas5384> i'm really excited about the new approaches you guys are developing
[01:34] <rick_h__> cool, glad ot hear it
[01:34] <sebas5384> congrats on that :D
[01:34] <sebas5384> talking about the machine view
[01:35] <sebas5384> when i imagined how it should be the gui for that
[01:36] <sebas5384> something like what we have in with app's bundles in our phones
[01:36] <sebas5384> but man, the work that it's being doing its awesome!! :)
[01:36] <rick_h__> sorry, not following 'apps bundles' phrasing
[01:36] <rick_h__> oh, you mean something like a box for each machine?
[01:36] <rick_h__> yea, I'll be very glad to get it out there for sure
[01:36] <sebas5384> yeah!, something like that
[01:37] <rick_h__> we looked hard at that, but the UX doesn't scale well
[01:37] <rick_h__> especially when you get into lxc/kvm containers inside machines
[01:37] <sebas5384> when you drag an app on top of another (in your phone) it's join al together in a bundle
[01:37] <rick_h__> and in 100's of machines those boxes get hard to manage, too big. No good way to go through them quickly/etc
[01:38] <rick_h__> yea, that's really kind of limited in size/scope
[01:38] <sebas5384> yeah I agree
[01:39] <rick_h__> but anyway our UX team spent many months going through design ideas and I think what we've got will be really nice
[01:39] <sebas5384> yeah for sure man!
[01:39] <rick_h__> a lot of idea left on the cutting room floor
[01:40] <sebas5384> yeah I can imagine
[01:40] <sebas5384> but its seems not be like the charm view, you know?
[01:41] <rick_h__> right, that's why there's a toggle up top. You have the service view, and the machine view
[01:41] <rick_h__> we'll work on adding other views in the future I think
[01:41] <sebas5384> great
[01:41] <sebas5384> so talking about this gui's things
[01:42] <sebas5384> we are studying about a Juju as a service
[01:43] <sebas5384> so for that you would have to manage a lot of environments, and some kind of a web gui for the juju quickstart tool
[01:43] <sebas5384> there are some other ideias like charm factory, etc... but juju as a service is a real need for us right now
[01:44] <sebas5384> there's plans to something like that?
[01:44] <rick_h__> interesting. it'd be interesting to get an idea of what you're looking for. What are the features/requirements and reasons for them.
[01:45] <sebas5384> rick_h__: sorry to throwing all this questions to you hehe
[01:45] <rick_h__> sebas5384: all good, happy to help/chat
[01:45] <sebas5384> rick_h__: :)
[01:45] <rick_h__> sebas5384: there's some work to go in this cycle, I think it was emailed to the -dev list after our vegas sprint?
[01:46] <sebas5384> really? didn't see it
[01:46]  * rick_h__ goes to check archive
[01:47] <sebas5384> I sow a guy from adobe asking for something like that
[01:47] <sebas5384> and then the conversation was focus in the api's I think
[01:48] <rick_h__> bah, can't find what I was thinking of
[01:49] <sebas5384> its ok :0
[01:49] <rick_h__> https://lists.ubuntu.com/archives/juju-dev/2014-May/002500.html is a little old thread that might be of interest. There's some work to help with the stuff you're asking about.
[01:49] <sebas5384> :)
[01:49]  * sebas5384 reading about
[01:49] <thumper> rick_h__: hey
[01:49] <thumper> whazzup?
[01:50] <rick_h__> yea, worth tracking if you have an interest in this stuff
[01:50] <rick_h__> thumper: have a doc for you to peek at and a bug to put into your brain
[01:50] <thumper> rick_h__: hmm... ok
[01:50] <rick_h__> thumper: not asking for anything today :)
[01:52] <sebas5384> rick_h__: I see, but my ideia is more for the gui
[01:53] <sebas5384> though the ideia of sharing envs is awesome
[01:54] <sebas5384> I have this vision to make so easy for people deploy and orchestrate their topologies as a drag and drop action
[01:54] <sebas5384> and juju-gui makes this happen!
[01:55] <sebas5384> so it should be easy to get a juju gui configured environment for you with some clicks and filling some inputs fields with the provider's info
[01:59] <thumper> davecheney: with you in 5 min, just need to get toast
[02:00] <rick_h__> sebas5384: right, but the GUI relies on the power of Juju to function to what you'll find is that we add basic support for ideas into juju and then grow on top of it with the GUI
[02:01] <rick_h__> sebas5384: but I'd love to get your use cases down to make sure we're thinking about them and addressing them. Maybe you've got ideas we've not thought about or some ideas we think are only kind of useful are very useful to you
[02:01] <rick_h__> sebas5384: so feel free to email the juju list, or me, or the gui list, and we can collect your input.
[02:01] <sebas5384> yeah for sure rick_h__ I love brainstorming
[02:02] <sebas5384> and iterating on top of feedback
[02:02] <sebas5384> is the more agile way to develop an app
[02:02] <rick_h__> but for now it's time for me to get a shower and head to bed. Night from the EST
[02:02] <sebas5384> avoids waste of features and focus in what it really adds value to the client
[02:03] <sebas5384> rick_h__: nice to talk to you, and thanks for your attention :)
[02:03] <sebas5384> rick_h__: good night to you man!
[02:04] <davecheney> thumper: ok
[02:04]  * thumper is eating quickly
[02:15] <davecheney> uniter_test.go:813:
[02:15] <davecheney>     c.Assert(result, gc.DeepEquals, params.StringsWatchResults{
[02:15] <davecheney>         Results: []params.StringsWatchResult{
[02:15] <davecheney>             {StringsWatcherId: "1", Changes: []string{
[02:15] <davecheney>                 firstAction.Id(),
[02:15] <davecheney>                 secondAction.Id(),
[02:15] <davecheney>             }},
[02:15] <davecheney>         },
[02:15] <davecheney>     })
[02:15] <davecheney> ... obtained params.StringsWatchResults = params.StringsWatchResults{Results:[]params.StringsWatchResult{params.StringsWatchResult{StringsWatcherId:"1", Changes:[]string{"wordpress/0_a_1", "wordpress/0_a_0"}, Error:<nil>}}}
[02:15] <davecheney> ... expected params.StringsWatchResults = params.StringsWatchResults{Results:[]params.StringsWatchResult{params.StringsWatchResult{StringsWatcherId:"1", Changes:[]string{"wordpress/0_a_0", "wordpress/0_a_1"}, Error:<nil>}}}
[02:15] <davecheney> [LOG] 0:00.471 INFO juju.provider.dummy reset environment
[02:15] <davecheney> [LOG] 0:00.471 INFO juju.state.apiserver [30] user-admin API connection terminated after 204.572645ms
[02:15] <davecheney> [LOG] 0:00.476 INFO juju.testing reset successfully reset admin password
[02:16] <davecheney> [LOG] 0:00.490 INFO juju.testing reset successfully reset admin password
[02:16] <davecheney> [LOG] 0:00.494 INFO juju.testing reset successfully reset admin password
[02:16] <davecheney> [LOG] 0:00.494 ERROR juju.state.apiserver.common error stopping *state.actionWatcher resource: state has been closed
[02:16] <davecheney> OOPS: 50 passed, 1 FAILED
[02:16] <davecheney> --- FAIL: Test (22.56s)
[02:16] <davecheney> FAIL
[02:16] <davecheney> FAIL    github.com/juju/juju/state/apiserver/uniter     22.721s
[02:21] <sebas5384> davecheney: pastebin dude! :)
[02:26] <davecheney> eeek
[03:10] <davecheney> menn0, thumper: nag for the OCR, https://github.com/juju/juju/pull/305
[03:10]  * thumper should check to see if he is OCR
[03:26] <thumper> davecheney: https://github.com/juju/juju/pull/290 if you feel like returning the favour
[04:23] <davecheney> thumper: i can do that
[04:23] <thumper> cheers
[04:31]  * thumper has another dependent branch ready
=== vladk|offline is now known as vladk
[06:35]  * davecheney deletes many duplicate worker/mockConfig types
=== rogpeppe is now known as rogpeppe1
[07:01] <jam> dimitern: morning, be there in just a sec
[07:02] <dimitern> jam, morning, me too :)
=== seelaman` is now known as seelaman
=== uru is now known as urulama
=== urulama is now known as uru-afk
[08:40] <TheMue> morning
[08:42] <dimitern> TheMue, morning
[09:00] <jam> morning TheMue
[10:04] <natefinch> jam: morning
[10:28] <natefinch> heh, gonna be a quiet day, since everyone from the US core devs except me gets a swap day today
[10:35] <TheMue> natefinch: so go to bed again
[10:35] <TheMue> natefinch: morning btw
[10:35] <TheMue> natefinch: ;)
[10:35] <natefinch> Heh, it's ok, maybe I'll get some work done today
[10:37] <jam> morning natefinch
[10:37] <jam> natefinch: I thought you wouldn't be in today, sorry I missed you
[10:37] <jam> (I have our standup in 10 min)
[10:37] <natefinch> jam: it's ok, no problem
[10:38] <natefinch> jam: I don't get a swap day today, since I didn't go anywhere :)  Just a regular week driving into the office.  It's funny, I forgot how much I dislike that :)
[10:40] <jam> natefinch: driving to work every day?
[10:40] <natefinch> yeah
[10:41] <natefinch> It's like 45 minutes in traffic, which is not the end of the world, but pleh.  What a waste of time.
[10:41] <natefinch> (The sprint was great, don't get me wrong, only the driving was a waste ;)
[10:43] <TheMue> natefinch: yes, and in case of having this all working days simply count up the lost time (beside the additional cost)
[10:46] <jam> dimitern: TheMue: vladk: standup?
[10:46] <TheMue> jam: already coming ;)
[10:46] <jam> natefinch: did the week go well other than the commute time ?
[10:47] <dimitern> jam, sorry, brt
[10:48] <natefinch> jam: The work week was great.  But at home, my A/C had a leak (yet again).  Such is life.
[12:59] <bodie_> morning all
[12:59] <TheMue> bodie_: morning
[13:00] <natefinch> morning bodie_
[13:18] <bodie_> addressed TheMue 's comments on PR 301, pushing them up in a moment here
[13:24] <bodie_> https://github.com/juju/juju/pull/301 should be good to go
[13:25] <bodie_> would appreciate LGTM!
[13:26] <TheMue> bodie_: I’m looking again.
[13:27] <TheMue> bodie_: It already has been ok, only the comments you now addressed
[13:27] <bodie_> :)
[13:27] <bodie_> TheMue, I mentioned this in the comments but fwereade had indicated he really wanted a refactor of the uniter context and runhook
[13:28] <bodie_> I'm working on a (pretty bloated by this point) PR to address a lot of that which splits hook context and hides a lot of content inside the new types
[13:28] <jam1> natefinch: just to remind you guys that I'm on
[13:28] <jam1> on-call IRC master today
[13:28] <jam1> so ping either jam1 or jam if you have questions
[13:29] <natefinch> jam1: thanks.  I think you can probably bail today.  Everyone except me has a swap day today.
[13:29] <natefinch> (everyone in the US that is)
[13:33] <TheMue> bodie_: LGTM
[13:33] <bodie_> TheMue, I still need another lgtm, right?  I'm never quite certain when I need two
[13:34] <TheMue> bodie_: one is ok
[13:34] <bodie_> cool, thanks!
[13:40] <jam1> natefinch: dang, I was hoping you meant I got a swap day, too :)
[13:40] <vladk> dimitern: I've committed https://github.com/juju/juju/pull/121
[13:42] <dimitern> vladk, thanks, I'll have a look
[13:42]  * TheMue too
[13:42] <dimitern> vladk, re your question about tests - using statetesting.AssertOneChange() and so should be enough, as these methods internally call StartSync
[13:43] <natefinch> jam1: haha
[13:58] <bac> allenap: in your vast azure experience did you ever look at using floating ips?
[14:07] <TheMue> dimitern: would you take a look at https://github.com/juju/juju/pull/306 too? I’m using now dynamic addreeses here.
[14:11] <dimitern> TheMue, sure, in a bit
[14:13] <TheMue> dimitern: thx
[14:14] <dimitern> TheMue, LGTM
[14:15] <TheMue> dimitern: ta (and jam1 too) ;)
[14:31] <dimitern> vladk, reviewed
[15:04] <dimitern> TheMue, please take a look https://github.com/juju/juju/pull/307
[15:04] <dimitern> when you can
[15:05] <TheMue> dimitern: sure, just seen the mail :)
[15:05] <dimitern> TheMue, cheers! :)
[15:05] <TheMue> dimitern: oh, ONLY 24 files :D
[15:06] <dimitern> TheMue, heh, changes to agent/environ config have this effect yes :)
[15:07] <TheMue> dimitern: yeah, some are that extensive
[15:08]  * dimitern needs to step out, bbl
[15:20] <alexisb> jog, welcome to juju-dev!
[15:20] <jog> thanks alexisb, happy to be here!
[15:21] <TheMue> jog: heya also by me (from Germany)
[15:24] <jog> hi TheMue, thanks
[15:31] <alexisb> natefinch, jam1: looks like we have a github defect open:
[15:31] <alexisb> https://github.com/juju/juju/issues/304
[15:31] <alexisb> rick_h__ pointed it out to me
[15:50] <TheMue> dimitern: if you’re back, you’ve got a review
[15:51] <bodie_> trying to understand a yaml error
[15:52] <bodie_> goyaml*
[15:52] <bodie_> "could not find expected directive name"
[15:52] <bodie_> as I understand it, directives are optional
[15:52] <TheMue> bodie_: where exactly you get it?
[15:53] <natefinch> alexisb: pleh, looks like those instance types were just added July 1
[15:53] <natefinch> (to AWS, not juju)
[15:53] <bodie_> http://paste.ubuntu.com/7794085/ TheMue
[15:54] <bodie_> TheMue, perhaps this is a different error -- it indicates the problem is on line 13
[15:54] <bodie_> which doesn't seem to exist
[15:54] <TheMue> *click*
[15:54] <TheMue> bodie_: could you also paste a code snippet?
[15:55] <bodie_> yeah... let me just shoot you the real thing highlighted on github
[15:55] <TheMue> ok
[16:01] <bodie_> TheMue, https://github.com/binary132/juju/blob/367f2b557fa09665bc77bff482e609ea9ac6be3b/worker/uniter/uniter_test.go#L267-L283
[16:01] <bodie_> that's where the yaml is getting generated
[16:01] <bodie_> I suspect the issue might be on line 279, but that's based on something that is known working
[16:02] <tasdomas> when trying to run 'juju debug-log' with trunk, I get: ERROR cannot open log file: open /var/log/juju/all-machines.log: no such file or directory
[16:03] <tasdomas> is there a way to view the juju log by ssh'ing into the state server?
[16:03] <bodie_> TheMue, perhaps looking at my spaghetti code tests isn't the best way to spend your time ;)
[16:06] <bodie_> http://paste.ubuntu.com/7794138/ will perhaps be more helpful (TheMue)
[16:07] <TheMue> tasdomas: juju ssh 0
[16:07] <TheMue> bodie_: hehe, will look there too
[16:07] <bodie_> the error is shown on line 1346 of that output
[16:07] <tasdomas> TheMue - that part I know, which log files do I look in?
[16:08] <TheMue> tasdomas: in that directory /var/log/juju take a look at machine-0.log
=== vladk is now known as vladk|offline
[16:11] <tasdomas> TheMue, thanks - that file seems to be not a complete replacement for the debug-log
[16:11] <TheMue> tasdomas: no, that’s the all-machines.log. debug-log only is the command for it
[16:11] <TheMue> tasdomas: local provider?
[16:12] <tasdomas> TheMue, ec2
[16:12] <TheMue> tasdomas: astonishing
[16:12] <tasdomas> TheMue - unrelated question - there is a way to connect to the juju mongo instance, right?
[16:13] <TheMue> tasdomas: in which way? with a cli client?
[16:13] <TheMue> bodie_: just for info, I’m still reading code and log
[16:13] <tasdomas> TheMue, yes, from within the state server
[16:14] <TheMue> tasdomas: afaik not, only using a coded client, like we do with mgo
[16:17] <tasdomas> TheMue, thanks
[16:17] <TheMue> tasdomas: sorry for having no better info
[16:18] <TheMue> tasdomas: do you have services deployed or only bootstrapped?
[16:18] <tasdomas> TheMue - I have services deployed, but I'm trying to debug an ec2 deployment
[16:20] <TheMue> tasdomas: the wanted all-machines.log is an aggregation of machne-0.log machine-1.log etc
[16:21] <tasdomas> TheMue, yes, I understand that
[16:22] <TheMue> tasdomas: that’s why I asked for deployed services. it’s strange you don’t find an all-machines.log then.
[16:26] <TheMue> bodie_: in your created output, line 11, why is outfile in square brackets?
[16:28] <bodie_> TheMue, it's my understanding that yaml permits lists in square brackets
[16:28] <bodie_> required is expected to be a list
[16:28] <bodie_> hmm
[16:29] <bodie_> I think the bit: content := fmt.Sprintf(actionsYamlFull, filepath.Base(actionsYamlPath))
[16:29] <bodie_> was the issue
[16:29] <bodie_> I've got it validating now
[16:31] <TheMue> bodie_: that’s not what I expect when getting a YAML prefixed error
[16:31] <bodie_> I'm not sure I understand the utility of that clause
[16:32] <TheMue> bodie_: I know YAML list written with - on new lines
=== vladk|offline is now known as vladk
=== vladk is now known as vladk|offline
[17:40] <jam> natefinch: fwiw T2 are going to be interesting, because they come with a lot of different constraints. They are cheaper, but must be in VPC (cannot be used in Classic mode), they "accrue" CPU credits that they can use to burst, etc
[17:40] <jam> They actually sound like a decent fit for the juju state server (if you consider that it usually goes quiet once everything is stable)
[17:40] <jam> but the networking around them means they aren't just like other instances
[17:42] <natefinch> jam: very interesting.  I hadn't looked into all the differences.... the CPU credits is kind of an interesting idea
=== vladk|offline is now known as vladk
[19:04] <allenap> bac: It rings a bell. It might all come back if you want me to look at something.
[19:06] <bac> allenap: looking into using floating ip on azure it seems to not be possible.  i was hoping to verify that with someone with lots of azure expertise.
[19:08] <bac> allenap: but we can chat about it tomorrow, since you're well past eod
[19:11] <bodie_> anyone know when fwereade will be around for me to pester?
[19:11] <bodie_> mgz will do as well
[19:46] <allenap> bac: I’m around now, but tomorrow would be better. Too many things to do this evening!
[19:47] <bac> allenap: sure
[19:47] <bac> tomorrow is better
[19:49] <bac> allenap: perhaps have a look at http://msdn.microsoft.com/en-us/library/azure/dn690120.aspx tomorrow and we can chat around 1300UTC
=== vladk is now known as vladk|offline
[20:40] <Egoist> Hi
[20:40] <Egoist> Is there any way to look how many units is in service, from hook code?
[21:50] <thumper> ha... thought it was quiet
[21:51] <thumper> had forgotten to start the irc client
[21:51] <ChrisW1> blah
[21:52] <thumper> o/ ChrisW1
[21:55] <ChrisW1> okay, so what does '0/' mean?
[21:55] <ChrisW1> o/, even
[21:56] <ChrisW1> seen it several times tonight, feeling out of touch
[22:20] <sebas5384> what would it be the better approach for charm scaling?
[22:22] <sebas5384> things like sharing files using nfs or glusterfs came to my mind
[22:24] <sebas5384> i found this storage charm but I don't know if is a good optin
[22:24] <sebas5384> option
[22:24] <sebas5384> marcoceppi: would you recommend the gluster-server and gluster-client charms for scaling?
[22:25] <voidspac_> wwitzel3: ping
[22:42] <sebas5384> ok, looking over ceph now
[22:58]  * thumper needs a git master
[22:58] <thumper> I need to unfuck a branch
[23:00] <cmars> thumper, mind if I join onyx standup today? also, I'm not too shabby at git
[23:00] <thumper> cmars: go ahead
[23:04] <ChrisW1> thumper: what's your git problem?