[13:08] <jcastro> aculich: hey I linked up with tim last week
[13:08] <jcastro> marcoceppi: do you have the juju go PPA handy?
[13:09] <davecheney> jcastro: ppa:juju/devel
[13:09] <jcastro> ta
[13:10]  * davecheney melts into the background
[13:10] <jcastro> hmm, so I've noticed we haven't written that down anywhere
[13:10] <davecheney> jcastro: mgz and davey are in charged of putting juju somewhere more official
[13:11] <jcastro> yeah, I'm writing the drafts for the getting started pages today
[13:11] <jcastro> so we can flip them over on thursdayish
[13:13] <marcoceppi> hum, I've been using the wrong PPA
[13:27] <evilnickveitch> jcastro - i'm doing a getting started as part of the go docs...
[13:27] <jcastro> evilnickveitch: ok, so what do I do with the page on the website?
[13:27] <jcastro> do we just remove it when newdocs lands?
[13:28] <evilnickveitch> jcastro - yes, thats what the thinking was, it should be able to go on Thursday, though the rest of the docs won't be finished
[13:28] <marcoceppi> jcastro: we could could just update the header link to go to the getting started page of the docs as the doc layout is nearly identical to the website atm
[13:29] <jcastro> that would be perfect!
[13:29] <jcastro> evilnickveitch: I think the only thing we should do is add a little snippet for python users.
[13:29] <jcastro> "Looking for older versions of Juju?" and then send them someplace.
[13:29] <evilnickveitch> hurrah!
[13:30] <evilnickveitch> heheh! yes, I can think
[13:30] <evilnickveitch> that would be a good idea. I will make sure to flag it up - i guess i can just point to the old docs (though that seems cruel)
[13:31] <evilnickveitch> maybe some sort of intermediary page
[13:31] <marcoceppi> jcastro: here's an earlier draft of the design, for reference: http://hostmar.co/canonical/go-juju-docs-go/getting-started.html
[13:31] <jcastro> evilnickveitch: well, if they're on python then i think giving them the old-non-changing docs would be the best bet
[13:31] <jcastro> at least for thursday
[13:32] <evilnickveitch> jcastro, agreed.
[13:34] <jcastro> man dude, that is looking _excellent_ btw
[13:37] <evilnickveitch> jcastro, thanks! marcoceppi has helped out a lot!
[13:38] <jcastro> we're preserving URLs right? Like with rewrites and whatnot?
[13:52] <jcastro> I take it the silence means we're burning the ships? :)
[13:54] <mgz> I think it means that no one knows who the we-s are in that context
[14:14] <marcoceppi> Does pyjuju work with raring on cloud providers?
[14:17] <jcastro> yeah
[14:17] <jcastro> well, it was for me last week
[14:18] <marcoceppi> must have just been a bad bootstrap
[14:22] <jcastro> man guys
[14:22] <jcastro> just followed the directions
[14:22] <jcastro> and deployed wp with juju-core with mysql
[14:22] <jcastro> Just Worked(tm)
[14:22] <jcastro> this will be awesome
[14:38] <ahasenack> 2013/04/22 14:38:23 ERROR agent died with no error
[14:38] <ahasenack> :)
[14:39] <jcastro> hah!
[14:39] <jcastro> gary_poster: hey, thanks for that tablet update last week
[14:40] <gary_poster> jcastro, np! Thanks for bug report.  Sounds like it was a great event
[14:41] <jcastro> hey so I think the only real problem I noticed is everytime m_3 went to deploy something the twitchiness of the click-to-hold + a touchpad gets a bit wonky
[14:41] <jcastro> I so want to right click on the box
[15:01] <m_3> jcastro: sounds about right... was hard to catch a right click without a move involved too
[15:02]  * m_3 chalks it up to coffee... not nerves ;)
[15:29] <FunnyLookinHat> Have you guys ever set up a wildcard subdomain redirect scenario for scaling?  I.e. have a single box redirecting / managing requests that need to be handled by specific boxes?  I can go into more detail, but don't want to waste your guys' time.  :)
[17:27] <jcastro> SpamapS: any objections to the categories: thing in metadata.yaml as policy?
[17:27] <jcastro> I know you don't like the entire idea. :)
[17:27] <SpamapS> jcastro: I am adamantly opposed to it.
[17:28] <SpamapS> its ridiculous and a waste of time. I wish you had brought it up at the summit so I could have lit it on fire :)
[17:29] <jcastro> ok well, you hate it, but the gui team needs them to make the GUI look nice for 13.04
[17:29] <jcastro> so, how can I make you at least not care about it? :)
[17:29] <SpamapS> jcastro: no they don't
[17:30] <SpamapS> jcastro: use tags. Weight the tags.
[17:30] <jcastro> I don't think they're doing that
[17:31] <SpamapS> jcastro: the reason I care about it is that categories are almost always poorly maintained in software
[17:32] <SpamapS> Its good to categorize flora/fauna, because you don't have whole new categories of flora/fauna every day
[17:33] <SpamapS> jcastro: go ahead and do it.. and while you're at it, tell yahoo to go back to categories too. :)
[17:33] <jcastro> ok
[17:33] <jcastro> well, as long as you don't hate me
[17:33] <jcastro> I don't like categories either
[17:34] <jcastro> but it won't matter
[17:34] <jcastro> people will use google for 95% of the pagehits anyway
[17:34] <SpamapS> I lurve you all mang
[17:34] <SpamapS> just think this idea is a time filling "because we can't think of anything interesting to do" waste
[17:34] <sarnold> which reminds me, is there any plan for the juju equivalent of apt-cache search?
[17:34] <SpamapS> sarnold: jitsu search
[17:35] <sarnold> (I haven't really -looked-, but I'd like to know the apt-cache search and apt-cache show equivalents..)
[17:35] <SpamapS> sarnold: hits jujucharms.com live tho
[17:35] <sarnold> SpamapS: if it saves me the effort...
[17:36] <sarnold> nice, .29s :) hehe
[17:37] <sarnold> thanks SpamapS :)
[17:44] <marcoceppi> There's charm search as well, but that runs against launchpad
[17:44] <jcastro> marcoceppi: branch proposed
[17:45] <jcastro> we need to sort out icons in policy too
[17:46] <jcastro> marcoceppi: you're putting the svg template thing in charm tools right?
[17:46] <jcastro> for `charm create`?
[17:47] <sarnold> marcoceppi: hrm, that didn't find the first one I searched for.. ('charm search powerdns' didn't return the same results that 'jitsu search powerdns' returned..)
[17:47] <marcoceppi> jcastro: yeah, I'll have a merge for that shortly
[17:47] <jcastro> we need to merge that into docs too
[17:47] <jcastro> but I am mulling waiting for newdocs
[17:47] <jcastro> then just have that be the first submission to that
[17:48] <marcoceppi> jcastro: or we could just tell evilnickveitch to slide it in as he writes them. Don't want to be too far out of sync. Though I guess that'd be a good first "test" merge
[17:49] <jcastro> sure
[17:50] <marcoceppi> jcastro: we should also document the workflow for submitting changes to the docs when newdocs land
[17:50] <jcastro> http://askubuntu.com/questions/52063/how-do-i-contribute-documentation-to-juju
[17:50] <jcastro> is what I use
[17:50] <marcoceppi> ah, perfect
[17:51] <jcastro> we need to snag the content from the mailing list wrt. that default image id stuff
[17:51] <evilnickveitch> marcoceppi, jcastro  What am I signed up for now?
[17:51] <jcastro> actually, doing that now
[17:51] <marcoceppi> jcastro: it's already listed as depreciated in the docs, just need to update AU and make sure it's removed going forward
[17:51] <jcastro> evilnickveitch: we need to add the icon stuff to policy, I sent you a mail just now
[17:52] <jcastro> marcoceppi: on it
[17:52] <evilnickveitch> okay, cool
[17:52] <evilnickveitch> I also have a "contributing" page that I have been working on that will be okay to go live at the same time
[17:53] <evilnickveitch> It details the use of the HTML tags I have made use of, etc
[17:54] <marcoceppi> cool
[17:54] <jcastro> perfect
[17:55] <jcastro> evilnickveitch: hey so what happens to things that you haven't converted yet?
[17:55] <jcastro> will you just pull in the pages as is?
[17:55] <jcastro> marcoceppi: hey so bkerensa approved my MP for the docs, lol
[17:57] <marcoceppi> jcastro: merged
[18:06] <jcastro> are you not entertained?
[18:06] <jcastro> what's next!
[18:08] <evilnickveitch> jcastro: hmmm, we'll wait and see on that. I am planning on having all the user guide stuff ready by Thursday
[18:08] <jcastro> mgz: which PPA will Juju 2.x live for 13.04?
[18:08] <jcastro> stable releases I mean, not devel, unless we're reusing ppa:/juju/devel
[18:09] <jcastro> jamespage: ^^ in case you know
[18:16] <mgz> jcastro: devel I would think
[18:17] <mgz> but we can work out if we want a stable-unstable one at the sprint in a few weeks
[18:17] <mgz> before actually getting 2.0 we don't need it
[18:18] <jcastro> ack
[18:39] <ahasenack> hi, does gojuju work with lxc yet? I don't remember
[18:40] <gary_poster> hey jcastro.  https://juju.ubuntu.com/docs/service-config.html (in "Creating charms") it says that config.yaml supports "str", "int" and "float" as types.  We also see "boolean" and "string" used, and we support them in the GUI.  (1) are there any other types? (2) The haproxy problem you saw was because we treat str values as text lines (no newlines) not text fields (with newlines)
[18:40] <jcastro> ahasenack: nope
[18:40] <ahasenack> jcastro: ok, thanks
[18:40] <jcastro> gary_poster: lots of sections of the docs are out of date
[18:40] <gary_poster> IOW the diff between <input type="text" and <text
[18:40] <jcastro> if you want to fix it and submit a mp that would be <3
[18:40] <gary_poster> <textarea
[18:41] <gary_poster> jcastro, ack, and would like to.  Right now we are asking in order to fix the ODS bug you filed though
[18:41] <jcastro> yeah so I have no clue, I usually read the docs heh.
[18:42] <jcastro> this seems like a hazmat question
[18:42] <gary_poster> jcastro, lol cool thanks
[19:18] <smoser> so https://codereview.appspot.com/8648047/ is merged in, right?
[19:18] <smoser> can i get someone to back that out?
[19:19] <smoser> i've been told by ubuntu foundations that this will not be a problem.
[19:37] <hazmat> gary_poster, only supported types are string, int, float, boolean
[19:37] <gary_poster> ack hazmat thanks, found in go code
[20:09] <jcastro> gary_poster: ack on having the series in the URL
[20:09]  * jcastro wonders if putting the version # in there would be better
[20:11] <gary_poster> jcastro, I think we need both
[20:11] <gary_poster> but could be wrong :-)
[20:11] <jcastro> in the url?
[20:12] <gary_poster> yeah, because isn't precise/wordpress-5 potentially different than raring/wordpress-5?
[20:12] <jcastro> I was thinking 12.04/wordpress
[20:12] <jcastro> and so on
[20:12] <gary_poster> oh
[20:12] <gary_poster> that might be better for being future-proof...
[20:13] <gary_poster> though in that vein maybe we should wait for the future to be hardened before we change that stuff
[20:13] <gary_poster> either way is fine with me
[20:14] <gary_poster> though less work is better than more work, in isolation from other considerations :-)
[20:15] <jcastro> I am for whatever gets us through to thursday. :)
[20:15] <gary_poster> lol +1
[20:15] <orospakr> hey, I've Stopped all of my EC2 VMs owned by juju except for the boostrap node/zookeeper.  I'd like to ask juju to ensure that every has been started that needs to be.
[20:16] <orospakr> s/that every/that every VM and service/
[20:17] <jcastro> so you killed the VMs after you deployed stuff?
[20:17] <smoser> who here can answer my query above ?
[20:17] <smoser> can we revert the cloud-init config change at https://codereview.appspot.com/8648047/
[20:18] <jcastro> smoser: I think core is in #juju-dev
[20:21] <orospakr> jcastro, I did. note that they are EBS backed and I shut them down cleanly.
[20:22] <jcastro> ok so you want juju to be able to just fire them back up?
[20:22] <orospakr> I'm worried that there are plenty of real circumstances that can cause cloud-hosted VMs to reboot, and as such, I'd like to test that.
[20:22] <orospakr> jcastro, well, that would be nice.  If I fire them back up myself, they get stuck in the agent down state.
[20:23] <orospakr> or at least, that's how juju status reports them, and the services they are meant to host appear inoperative.
[20:23] <orospakr> is it possible that I'm asking the wrong questions?
[20:23] <jcastro> questions seem fine to me
[20:23] <jcastro> I am just not sure how juju agents handle being shutdown
[20:23] <jcastro> hazmat: ?
[20:24] <marcoceppi> jcastro: I would have thought it was added to upstart, but that might not be the case.
[20:25] <orospakr> my general goal is to be sure that I can bring the entire system back up from virtually cold (assuming the EBS volumes remain intact).
[20:25] <jcastro> yeah, this might be a better question for the list I think, I think the people who know for sure aren't here right now
[20:25] <orospakr> alright, I'll start one up manually, and rummage around via ssh to see if everything obvious is running.
[20:26] <hazmat> orospakr, its not really supported at the moment because the address changes when the instances are started dont ripple through the graph
[20:26] <hazmat> ie updating relation addresses
[20:26] <orospakr> hum. okay. so, generally, a restart really should involve a redeploy?
[20:27] <orospakr> for instances hosting services that are effectively stateless, like application servers, that's not a big deal. for instances hosting more persistent things, like postgres and mongo, I'm keen to have a better story. Is the final answer just automatic backups?
[20:29] <orospakr> fwiw, I just confirmed that jujud does come up on my instances when I manually restart them.
[20:29] <orospakr> they remain agent-down, however.
[20:30] <orospakr> (according to bootstrap node/juju status)
[20:31] <hazmat> orospakr, final answer atm is not supported, long term instances should re-record their address and propogate to their relations
[20:34] <orospakr> hazmat, OK. is there a best practice to avoid losing data and bringing platform back up quickly after solar flares happen to, say, knock out EC2 East Zone for a while?
[20:34] <FunnyLookinHat> Is there any way for me to deploy multiple "instances" of a service to a single VPS? i.e. let's say I have a simple PHP app that manages a to-do list and I want to deploy a copy of it for each user I have on the fly - can that be currently accomplished with Juju ?
[20:34] <FunnyLookinHat> As far as I can see right now - each "instance" gets it's own VPS
[20:36] <sarnold> FunnyLookinHat: that's a long-term goal for the golang juju implementation; you can bodge it together using the jitsu 'deploy-to' command, but that isn't exactly a supported mechanism
[20:37] <FunnyLookinHat> sarnold, Right right - I read up on that... if I were to use deploy-to I'd have to put some fancy into my charms to handle that as well, correct?  Otherwise they'd probably just go ahead and overwrite themselves.
[20:38] <sarnold> FunnyLookinHat: yes, you'd need to write the charms to be careful; that's probably going to be required of the eventual golang supported solution too
[20:38] <FunnyLookinHat> sarnold, Ok great - thanks :)
[20:54] <hazmat> orospakr, atm its manual snapshot and cross region snapshot copy
[20:54] <hazmat> orospakr, i've been putting together  a suite of aws services exposed as charms.. rds, elb, route53
[20:54] <hazmat> was just working on an aws-snapshot one
[20:55] <hazmat> the other ones are @ http://jujucharms.com/search?search_text=aws
[20:57] <orospakr> hazmat, well, the restoring from snapshots still would seem to suffer from the issue you described before: address changes not properly re-propagating after a restart.
[20:59] <hazmat> orospakr, yup
[20:59] <hazmat> orospakr, but the data is around as a sanity check
[21:00] <hazmat> orospakr, volume management is slated for juju for13.10 at which point hopefully this will get better. i expect we'll have a public roadmap out mid may. all the devs has been working hard to get the new version/rewrite out the door
[21:02] <orospakr> oh wow, cool.
[21:02] <orospakr> well, thank you very much. :)
[21:17] <hazmat> FunnyLookinHat, -force-machine to force non conflicting services to the same machine in go juju.. in pyjuju its available as the jitsu deploy-to add on
[21:20] <thumper> morning hazmat
[21:21] <hazmat> thumper, greetings
[21:43] <b1tbkt> did a 'destroy-environment' and released all nodes from maas (at which point they showed 'Ready). subsequent juju deploy is now just hanging at 'connecting to environment' after four hours.
[21:43] <b1tbkt> do I need to delete ssh keys somewhere, maybe?
[21:49] <b1tbkt> nm...system console was waiting for me to see its complaint about pre-existing lvm on its disk
[23:39] <mwhudson> does juju-go work on arm?
[23:41] <davecheney> mwhudson: it should
[23:41] <davecheney> i've tested the client a few times on arm a while back
[23:41] <davecheney> there are no problems compiling it
[23:42] <mwhudson> davecheney: ok
[23:42] <mwhudson> davecheney: https://launchpad.net/~juju/+archive/devel doesn'
[23:42] <mwhudson> davecheney: https://launchpad.net/~juju/+archive/devel doesn't appear to have arm builds
[23:43] <davecheney> mwhudson: it will for the next build
[23:43] <davecheney> it wasn't a priority before
[23:43] <mwhudson> cool
[23:43] <mwhudson> thanks
[23:43] <davecheney> if you are running armhf you can build from source
[23:44] <mwhudson> yeah
[23:44] <davecheney> if you are running raring, you might even be able to run all the unit tests
[23:44] <mwhudson> i might just wait 24 hours :)
[23:44] <davecheney> mwhudson: probably a better move :)