_mup_Bug #865163 was filed: default-series option has surprising behaviour <juju:New> < https://launchpad.net/bugs/865163 >09:16
niemeyerGood mornings!12:46
rogpeppeniemeyer: hiya!12:56
fwereadeheya niemeyer13:00
niemeyerHey folks!13:02
* niemeyer grabs coffee before engaging into juju awesomeness13:03
niemeyerhazmat: Yo13:23
hazmatniemeyer, greeting13:32
niemeyerhazmat: Alan seemed pleasantly surprised with juju13:33
hazmatniemeyer, awesome!13:34
hazmatniemeyer,  they've been setting up a saas system with plone.. ploud.. i'm glad you guys had a chance to talk13:34
niemeyerhazmat: Spent good part of Saturday's afternoon hacking a django "platform" charm with Sidnei.. very cool stuff13:35
hazmatniemeyer, sweet!13:35
niemeyerhazmat: and he was around for part of it13:35
hazmatniemeyer, did you guys use bash? or python?13:35
niemeyerhazmat: Bash for the low level charm integration.. but it's a surprisingly small amount of code13:36
niemeyerhazmat: Checks the branch out from Launchpad, and puts it live13:36
hazmatniemeyer, indeed, checking out revision control is the only main thing i used python for13:36
hazmaton a wsgi charm13:36
hazmatniemeyer, virtualenv for requirements.. but figuring out the db dependency from config was a little odd13:37
hazmatmade me want runtime dependency declarations13:37
niemeyerhazmat: Ah, interesting.. we didn't go so far13:37
niemeyerhazmat: We were thinking about a python-level or app-level hook for installing dependencies13:38
hazmatniemeyer, yeah.. as it is right now i've got optional relations for lots of things, and i just populate a conf file with the properties for the app13:38
niemeyerhazmat: Ah, sweet!13:38
hazmatusing gunicorn for serving up the app ( easy multi-process using a pre-fork model)13:38
niemeyerhazmat: We were going in the direction of an nginx server13:38
hazmatniemeyer, nginx is a good front end, but i still expect that to be a separate unit.. within a unit the socket bind, fork, saves an extra hop13:40
niemeyerhazmat: Hadn't heard of gunicorn before13:40
hazmatwhich is needed for the frontend anyways13:40
hazmatniemeyer, its a nice front end to wsgi apps supports several options.. gevent, pre-fork workers, etc.13:41
niemeyerhazmat: Nice, we should talk to sidnei about this13:41
hazmatniemeyer, sounds good13:44
hazmatniemeyer, just a heads up... i'm going to be working on solving transient zk disconnect issues and upstartifying things this week14:22
niemeyerhazmat: Good timing.. on a call with fwereade and we were just talking about that :-)14:23
hazmatfwereade, niemeyer, cool, how's the remote repo working coming?14:23
niemeyerhazmat: Wanna join?14:24
hazmatniemeyer, definitely14:24
niemeyerhazmat: sent14:24
robbiewniemeyer: fyi, I updated the call to use my conference number14:58
niemeyerrobbiew: Joining15:01
niemeyerLunch time!16:05
robbiewFYI -> https://juju.ubuntu.com/Testing16:07
SpamapSlol.. look at the example they use for server templates. ;)16:32
SpamapS"A server template could be used, for example, to build a server containing a pre-installed WordPress system and database"16:32
niemeyerrobbiew: Wow!16:50
niemeyerrobbiew: Integrates very neatly16:50
* robbiew is so happy IS added the rawhtml option to our wiki ;)16:51
niemeyerrobbiew: +1 :)16:53
m_3robbiew: rawhtml... yay16:53
* rogpeppe is off. see ya tomorrow.17:00
robbiewrogpeppe: cya17:09
niemeyerrogpeppe: Cheers! On a call now, but will review your branches before you're back :)17:16
_mup_Bug #865550 was filed: Provide automation for including apparmor profiles in charms <juju:New> < https://launchpad.net/bugs/865550 >17:35
niemeyerFour hours of meetings should be good enough for a Monday18:02
* niemeyer tries to actually review some code now18:02
_mup_juju/unit-info-cli r423 committed by kapil.thangavelu@canonical.com19:15
_mup_copy host resolv.conf to container before attempting any resolutions, in the customize chroot, this was being masked by the local package cache19:15
* hazmat heads out to doctor appt, bbiab19:18
niemeyerhazmat: Good luck there! :)19:26
niemeyerhazmat: When you're back, would like to ping you on a minor19:26
fwereadeniemeyer: ping19:32
niemeyerfwereade: Yo19:32
fwereadethanks for the review19:32
fwereadeniemeyer: one I'm about to address and merge, one addressed in a followup, one I intend to address in a followup19:32
niemeyerfwereade: Cool19:33
niemeyerfwereade: Branches seem to be missing pre-reqs, btw19:33
fwereadereally? crap, let me check19:33
niemeyerfwereade: The require-default-series one is.. hmm.. interesting19:33
niemeyerfwereade: Yeah, it's fine as I can imagine what the stack is19:33
niemeyerfwereade: But it's good to have that in place in general19:34
fwereadeniemeyer: consider that one speculative, I wasn't sure you'd like it but itseemed better to implement while you were away than to moon around wondering19:34
fwereadeniemeyer: definitely19:34
niemeyerfwereade: It's good speculation.. it seemed wrong and I was tempted to ask for more information19:35
fwereadeniemeyer: yeah, looks like I've missed at least one :( sorry19:35
niemeyerfwereade: But pondering on the problem for a while, you have a point19:35
fwereadeniemeyer: it's clint's point really :)19:35
niemeyerfwereade: _Today_, detecting the revision will yield surprising results more often than it will be useful19:36
niemeyerfwereade: Well, SpamapS has a point then :-)19:36
fwereadeniemeyer: detecting the series, right?19:36
niemeyerfwereade: Yeah, sorry19:37
niemeyerfwereade: Detecting is really the long term solution19:37
niemeyerfwereade: But, there are two points that make it suboptimal:19:37
niemeyer1) There's a single series that work today19:37
niemeyer2) We do auto-updating of the environment config, which can yield surprising results19:38
fwereadeniemeyer: 3) we want people to be able to run juju from non-ubuntu systems19:39
fwereadeniemeyer: I do feel it would be nice to be smarter about it, but I think I've convinced myself at least that Least Surprise is the right principle to follow at this point19:40
fwereadeniemeyer: the fact that auto-updating can *still* break this solution is (I think) a point against auto-updating, rather than against the solution19:40
niemeyerfwereade: 3) is kind of irrelevant in this specific case19:41
fwereadeniemeyer: feels somewhat relevant to the long-term case, but I don't really feel that specific point is going to be a fruitful avenue of discussion right now ;)19:42
niemeyerfwereade: The fact someone wants to be able to work in environment B isn't good reasoning for not doing something in environment A which is actually more common right now19:42
fwereadeniemeyer: granted19:43
niemeyerfwereade: Yeah, least surprise is good too19:43
niemeyerfwereade: and it's actually exactly what we were going at with the original approach, though19:43
niemeyerfwereade: If you're siting on a machine running Natty, doing things remotely with Natty is expected19:44
niemeyerfwereade: The problem, and the reason why it has my sympathy, are points 1) and 2) above19:44
niemeyerfwereade: For 1), natty doesn't exist19:44
niemeyer(for juju)19:44
fwereadeniemeyer: well, indeed :)19:44
niemeyerfwereade: re. 2), auto-updating means we'd get different behavior on an _existing_ environment post-bootstrap19:45
niemeyerfwereade: Which is quite awkward19:45
fwereadeniemeyer: yep19:45
niemeyerfwereade: I'm happy to move forward with the simplistic way for now..19:47
niemeyerfwereade: We'll have to change this once there are other releases, but we have some time to think19:47
niemeyerfwereade: I'll just check it up with hazmat19:47
fwereadeniemeyer: cool -- I agree it's not perfect, but it feels like a sensible short-term solution19:47
niemeyerfwereade: +119:48
niemeyerfwereade: Awesome.. just reviewing docs, but I think we're settled as far as our conversation goes19:48
niemeyerfwereade: We need the auto-revision bumping now, and the fake testing server19:48
niemeyerfwereade: Probably in that order, so that we can get things rolling into the Ubuntu front while we sort out the testing and all19:49
niemeyerfwereade: WDYT?19:49
fwereadeyep, sounds good to me19:53
fwereadeniemeyer, in case you didn't see, sounds good to me19:55
niemeyerfwereade: Cool!19:55
niemeyerfwereade: Thanks for checking out19:55
niemeyerfwereade: I'll finish reviewing the doc branch, but we're in sync I think19:56
fwereadeniemeyer: cool19:56
fwereadeniemeyer: there will be another doc tweak at some stage, to fix up the draft I mention19:56
fwereadeniemeyer: but, well, it is a draft :)19:56
niemeyerfwereade: Yeah, cool :-)19:57
niemeyerAram: Hey!20:33
niemeyerhazmat: Can you please check out this when you have a moment: https://code.launchpad.net/~fwereade/juju/require-default-series/+merge/7787820:50
niemeyerI'm going to step out for some exercising and be back later today20:50
fwereadeniemeyer: (I've assumed it was; the other place is the new-user tutorial, where every extra character counts, IMO)20:53
fwereadeniemeyer: (and now I've convinced myself I'm wrong: the full charm url will show up in status, and it'll be clearer that the two places reference the same thing if they look like the same thing)20:59
* hazmat checks out branch22:13
niemeyerfwereade: I'm here, but you're probably not.. :p23:28
SpamapSniemeyer: so, how is the eureka push going?23:33
SpamapSI've been distracted with other Ubuntu stuff all day23:33
niemeyerSpamapS: Very well.. the only critical bit is the client-side store work23:34
niemeyerSpamapS: fwereade's branches are in pretty good shape, though.. we'll probably have everything in by tomorrow, or wednesday the latest23:34
niemeyerSpamapS: fwereade will then work on a fake server the rest of the week so we can be sure that the store support _actually_ works23:35
niemeyerSpamapS: and then finish the real store in the next couple of weeks23:35
SpamapSWe can always release an updated client if the API has to change23:36
niemeyerSpamapS: That's cool, but at least the that work we're finishing tomorrow/wednesday should really be in23:37
niemeyerAram: ping23:38
Aramniemeyer: pong.23:39
niemeyerAram: Hey!23:39
SpamapSniemeyer: indeed, its a bit more of an abuse of the process to SRU a whole new feature in than it is to just SRU a new API version in. :)23:40
niemeyerSpamapS: Yeah, and also because my _hope_ is that we'll have zero incompatible changes in the 11.10=>12.04 time frame23:40
SpamapS+1 from me on that23:41
niemeyerSpamapS: So would really like to have the store work now, so that if we have to SRU changes they use the same user interface23:41
SpamapSwould go a long way to building up user support if they're able to smoothly transition to the PPA and/or 12.04 without having to do anything to their charms/running envs/etc.23:42
niemeyerSpamapS: Exactly23:42
niemeyerSpamapS: That's my hope really23:42
SpamapSSeems like the last of the big structural local storage changes are in with juju-origin and the local charm repository organization.23:42
niemeyerSpamapS: juju-origin is kind of minor in that regard23:43
SpamapSI do think we'll have to see some backflips in code as the ZK topology changes..23:43
niemeyerSpamapS: the charm url user interface is the big deal23:43
niemeyerhazmat: ping23:43
niemeyerSpamapS: another bit we have to sort out is being able to tag formulas as incompatible with a given revision in a nice way23:44
niemeyerSpamapS: But that seems fine for an SRU23:45
SpamapSthats another one where it fits nicely the SRU desire to make sure interoperability with remote services is maintained23:45
hazmatniemeyer, pong23:46
niemeyerhazmat: Yo!23:47
niemeyerhazmat: How're things going there?  Still churning, or more like heading to a beer? :)23:48
hazmatniemeyer, really hoping to have the unit get cli and addresses in relations by default in for oneiric23:48
niemeyerhazmat: Neat23:48
hazmatniemeyer, both in the queue23:48
niemeyerhazmat: I'll check that out today still23:48
SpamapShazmat: that would be *saweet*23:48
niemeyerhazmat: Awesome23:48
hazmatniemeyer, i'm still running into occasional issues with the local provider stuff, think it might be because i switch local networks so often23:49
niemeyerhazmat: Hmm23:49
hazmatnot sure, still also seeing an occasional lxc problem when i try to ssh into a machine23:49
hazmati committed a fix for the network issue, it always manifests as xmlrpc.launchpad.net temp failure in resolution23:49
SpamapSI'll gladly do a mass bug file/fix in the charm repo once that is done, as that gives us much better reliability in non-ec2 environments23:49
hazmatnot sure if its a good fix though23:49
hazmatworks for me atm though23:50
hazmatpty allocation error re lxc problem23:50
hazmatonly a reboot fixes it once it shows up23:50
hazmatniemeyer, having a look over fwereade's branches atm, still churning23:50
hazmatand that's not mutually exclusive to a beer ;-)23:50
hazmattdd ftw ;-)23:50
hazmatsaw the new db pkg for go23:51
hazmatelmo was at the surge conference saw him the last day we chatted a bit23:51
niemeyerhazmat: :-)23:51
niemeyerhazmat: On that last branch, check-latest-formulas, I think we need something like23:52
niemeyer /charm-info?charms=<urls>23:52
niemeyerSo that we can get metadata + bundle-sha256 at once23:52
niemeyerfor all the charms23:52
hazmatniemeyer, sounds good, as we learned rest purisms fails for efficiency23:53
niemeyerIt feels bad to do _3_ requests per charm download23:53

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!