/srv/irclogs.ubuntu.com/2013/11/14/#juju-gui.txt

evilnickveitcharosales, hatch, gary_poster  thanks !00:22
arosaleswe'll catch you tomorrow evilnick00:46
hatchrick_h_: looks like parallels 9 supports 13.10 so i'll need to upgrade if I want to use it on there...just fyi01:27
=== gary_poster is now known as gary_poster|away
rick_h_hatch: coolio03:36
=== gary_poster|away is now known as gary_poster
gary_posterhey frankban.  thanks for charm branch.  sorry again for bad qa.  trying out your branch11:01
frankbangary_poster: I was going to reply to your email, it does not seem a bad qa IMHO, it seems just a network connection problem, the environment setup exited with an error11:01
gary_posteroh!11:01
frankbangary_poster: so the charm is good, the less good thing is that when setup exits with an error we should call setup again when the user runs, e.g. make unittest11:02
frankbangary_poster: so the branch I proposed is just a fix for that (remove the activate file if setup fails, so that next time setup is called, pip install is re-executed))11:03
gary_posterfrankban, oh, so when no dists were found for lazr.authentication for some reason, which I missed because I was moving too fast, then everything above it (yaml and mock) were effetively aborted?11:04
frankbangary_poster: yes, I checked with a fresh lxc, and the dependencies are all correctly installed (I just found the missing rsync sysdep, added in the branch). so it seems pip install -r is transactional, so you were starting the test suite without any dependencies11:06
frankbangary_poster: also proposed a similar quickstart branch. the logic si the same: "Remove the activate file if the requirements installation process fails."11:07
gary_postercool11:09
gary_postertrying charm build again11:09
frankbangary_poster: so good news, the charm does not introduce bugs in the testing infrastructure. bad news: PIP_DOWNLOAD_CACHE prevented my to find this problem earlier11:09
gary_posterfrankban, ack.  Don't see a way around that unless we make the download cache shared. what do you think of having a PIP_DOWNLOAD_CACHE as a branch that the makefile checks out?11:13
gary_poster(or bzr ups)11:13
gary_posterfrankban, lazr.authentication reliably fails for me. :-( http://pastebin.ubuntu.com/6415348/ investigating11:16
frankbangary_poster: with the branch I proposed, at least "pip install -r ..." is retried in case of previous errors. So, for example, the first "make" failed yesterday for you, and the subsequent "make test", instead of showing import errors, retries the requirements download.11:16
gary_posterack, definite improvement11:16
gary_posterinteresting.  https://pypi.python.org/pypi/lazr.authentication is only owned by leonard.  he didn't transfer. but...yeah.  I can't downoad that, for some reason.11:22
frankbangary_poster: here is the full verbose log of "pip install lazr.authentication" without cache in my local machine, if it can help: http://pastebin.ubuntu.com/6415408/11:33
frankbangary_poster: generated using "pip install -v --log lazrauth.log lazr.authentication"11:34
gary_posterack thx looking at charm pip log too11:34
gary_posterfrankban, charm log excerpt very diff: http://pastebin.ubuntu.com/6415421/11:36
gary_posterCould not fetch URL https://launchpad.net/lazr.authentication (from https://pypi.python.org/simple/lazr.authentication/): <urlopen error [Errno -2] Name or service not known>11:37
gary_poster!11:37
gary_posterhttps://launchpad.net/lazr.authentication loads fine in browser11:37
frankbanCould not fetch URL https://launchpad.net/lazr.authentication/+download (from https://pypi.python.org/simple/lazr.authentication/): <urlopen error [Errno -2] Name or service not known>11:39
frankbanoh, yeah11:39
frankbangary_poster: the same for https://launchpad.net/lazr.authentication/+download I guess11:40
gary_posterfrankban, yes. and http://pastebin.ubuntu.com/6415433/11:41
frankbanhum...11:42
gary_posterand http://pastebin.ubuntu.com/6415437/11:42
gary_posterfrankban, what do you think of having a PIP_DOWNLOAD_CACHE as a branch that the makefile checks out, or ups?11:49
gary_posterlp:~juju-gui/juju-gui/charm-download-cache or something.11:50
gary_posterfrankban, bizarrely, worked that time. :-/11:53
gary_posterstill think download cache would be a good idea11:54
frankbangary_poster: huh? does it work now? yes I agree re the separate branch, medium priority? It's just for the tests and I'd like to understand why lazr.auth failed for you11:56
gary_posterfrankban, yes it works now.  no idea why.  it can now access those pages when it could not before. :-/11:58
gary_posterfrankban, separately https://codereview.appspot.com/26530043/LGTM11:58
gary_posterhttps://codereview.appspot.com/26530043/ LGTM11:58
gary_poster(with comment)11:58
frankbangary_poster: confused about the comment, how do you use $ACTIVATE to diagnose?12:00
gary_posterfrankban, you can source the file and then use the venv's python to see if you can dupe the problem12:00
frankbangary_poster: aha! good suggestion, changing also the other quickstart branch then12:01
gary_postercool12:01
gary_posterfrankban, https://codereview.appspot.com/26540043/ also LGTM with trivial12:04
frankbangary_poster: great, thank you12:06
gary_posterwelcome frankban.  replying to my own email about this.12:06
frankbangary_poster: both branches landed, thanks for the reviews and for the email! 12:38
frankbantrying to build new quickstart packages now12:39
=== TheRealMue is now known as TheMue
gary_posterrick_h_, will ping soon, sorry, still talking13:36
rick_h_gary_poster: np13:36
jcastrofrankban, heya, so what's the plan for putting quickstart in the normal juju ppa?13:42
gary_posterbenji, will be late sorry.  rick_h_ joining now13:57
benjigary_poster: no worries13:58
jcastrorick_h_, hey if you have time today, when the stuff lands in prod on jujucharms.com I'd like to write it up, maybe talk me through it over g+ later?14:05
rick_h_jcastro: rgr14:06
frankbanjcastro: not sure about the plan, gary_poster can give a better answer when he's free from calls14:23
* jcastro nods14:23
bachi frankban, can you give me some advice on how to use and test @gen.coroutine?14:26
frankbanbac: 1min14:26
bacfrankban: grab this branch when you get a chance lp:~bac/charms/precise/juju-gui/increment-deployments14:31
frankbanbac: looking14:49
bacfrankban: clearly i'm not returning the values correctly.14:54
frankbanbac: gen.Return is something you have to raise14:54
bacfrankban: and then is it a real exception or is that a return mechanism?14:55
frankbanbac: when you have a coroutine, you can either exit without values (without returning or just with "return" without values) or you can raise a gen.Return, that will be treated as a return value. This is a workaround to python2 not being able to use return in a generator.14:56
gary_posterjujugui, I need helpfrom someone who can test the charm upgrade.  I was going to try and do it but am swamped in calls.  You would set up charm r76 the way IS has it (https://wiki.canonical.com/InformationInfrastructure/WebOps/CDO/JujuGui) and then upgrade to the newest version of the charm and document what needs to be done14:57
frankbanbac: in general, when you have a function decorated with gen.coroutine, the test is easier to write using the tornado.testing.gen_test decorator.14:57
gary_posterfor webops14:57
gary_posterbac or benji, I can talk now with one of you, and move the other to the afternoon.  preference?14:58
bacgary_poster: same/same14:58
gary_posterack14:58
bacgary_poster: so now or later?  :)14:58
gary_posterbac, benji's not responding, so let's do it now14:59
bacfrankban: ok, i'll look for gen_test14:59
benjigary_poster: that's fine, we can go this afternoon14:59
gary_posterthanks benji15:00
benjigary_poster: but I do need a quick pre-imp on a card after you're done though15:00
gary_posterbenji heh ok.  not a lot of time.  will squeeze in15:00
frankbanbac: the idea with gen_test is that the test itself can use yield in order give the execution control back to the ioloop and wait for the result15:01
bacfrankban: ok, on a call with gary_poster atm.  be back shortly.15:03
frankbangary_poster: I'll test the charm upgrade. So bzr revno 76 of lp:charms/juju-gui, juju-core environment and builtin-server=false, correct?15:10
gary_posterthank you frankban !  yes15:10
frankbancool15:10
gary_posterfrankban, also note the ga-key instructions on that page15:10
frankbangary_poster: oh, so currently we are not using ga... ok so the new options are ga and builtin-server15:12
gary_posteryes15:12
frankbancool15:12
hatchDart 1.0 was released today http://news.dartlang.org/2013/11/dart-10-stable-sdk-for-structured-web.html15:21
hatchso now we can port to Dart?15:22
* hatch runs15:22
hatchMakyo: OneTab needs the ability to 'store' a single/group of tabs15:25
hatchIt says it supports that but I can't figure out how :)15:26
hatchjujugui can I search commits on launchpad? I'm trying to find the commit to juju-core which added the unit details object15:31
bachatch: much easier with bzr locally, no?15:33
frankbangary_poster: the old charm downloaded the gui release file v0.10.1. Is this surprising or am I missing something. I expected the old charm to download the last release. oh, maybe it was the last to be tgz?15:33
hatchbac: it's looking like it15:33
frankbanyes it was15:34
bachatch: do you know what file?15:34
hatchbac: well it was a series of them15:35
bachatch:  you can do 'bzr log <path-to-file>' and it'll only show you commits that changed that file15:35
hatchif I could find the bug that would help either15:35
gary_posterfrankban, yeah probably last tgz15:35
frankbangary_poster: do I have to simulate latest version, or just proceed with the upgrade?15:35
frankbangary_poster: maybe it's better to just go on. let's see15:36
gary_posterfrankban, I think proceed is fine15:36
frankbanok cool15:36
hatchfound it https://bugs.launchpad.net/juju-core/+bug/122456815:37
_mup_Bug #1224568: Improve hook error reporting <juju-core:Fix Released by themue> <https://launchpad.net/bugs/1224568>15:37
hatchjeesh that took forever15:37
Makyojujugui call in 10.15:50
frankbangary_poster: so revno76 does not work in juju-core (apache returns a 403, I guess it is related to the fact the gui was served from inside the charm directory.15:51
gary_posterfrankban, so what they told us must be wrong?15:51
frankbangary_poster: they either fixed this issue or used another, less restrictive, version of juju-core15:52
frankbangary_poster: juju-core is another viarable now, but I remember we had a fix for the permission problem...15:53
frankbanvariable even15:53
gary_posterwe did15:53
gary_posterfrankban, you saw #webops?15:55
frankbangary_poster: yes, and confirmed, solved giving +ar to /var/lib/juju/agents/unit-juju-gui-0/charm15:55
gary_posterack15:55
frankbangary_poster: so I think it worked at the time with an older juju-core, but then they introduced this restriction15:56
gary_posterok15:56
gary_posterthat rings a bell15:56
Makyojujugui call in 215:58
gary_posterrick_h_, pingy ping16:00
adeuringbenji: could you have another look at my MP? The changes are big enough to need another review...16:04
benjiadeuring: sure (I'm on a call now, but will as soon as I can)16:05
adeuringthanks!16:05
hatchfrankban: is there a way I can inspect the web socket traffic from the juju-core side?16:13
hatchahah I finally got it sending the information16:16
frankbanhatch: if you bootstrap with --debug all the ws traffic is in the debug-log16:18
rick_h_hatch: or Makyo can I get one of you to do a first pass reviews on https://codereview.appspot.com/26290043/ w/o QA and just one for now. I've got to qa it all in a live env and I'll write up QA instructions once I make sure it all works right. Then I'll bug for two real reviews and a QA. 16:19
MakyoSure16:19
hatchI'll take a quick glance16:19
rick_h_Makyo: thanks16:20
hatchoh Makyo you're on it16:20
bachey frankban, i want to mock AsyncHttpClient.fetch() so that it returns a response with a non-200 code.  what thing should i actually return?  a Future?16:20
hatchrick_h_: don't like composition hey? ;)16:21
rick_h_hatch: what do you mean? 16:21
hatchI just had app.js open, you removed the mixin in favour of a utils file16:21
rick_h_hatch: oh, I don't like tools that don't need any access to object state and require building into a dummy object just to test :P16:21
frankbanbac: yes, a concurrent.futures.Future (with a result already set)16:21
rick_h_hatch: right, but it meant that it was spread across views/utils, app, the extension, the topology, etc16:22
hatchyeah that was all over the place16:22
rick_h_hatch: but yea, I felt this cleans things up as I needed to be able to wrap all places that deploy a bundle into one path I could test and add a watcher to16:23
rick_h_and an extension didn't get me anything16:23
hatchyeah true true16:24
Makyorick_h_, yeah, first pass looks good.16:30
rick_h_Makyo: cool, appreciate making sure there's nothing crazy while I get the QA stuff in order. 16:30
hatchrick_h_: do websocket frames wrap for you? It looks like it's broken in chrome16:31
hatchI'm asking because you're on dev :)16:32
rick_h_hatch: I'll have to look in a few. My charm is in progrerss of re-working due to changing the source 16:32
hatchsure no rush https://code.google.com/p/chromium/issues/detail?id=286419 claims it was fixed in chromium but it would be too bad if that wasn't back ported16:33
hatchahh looks like they just have some bad css in the inspector now16:35
hatchI wonder if I can provide a patch...16:35
benjiadeuring: I've finished looking at the branch.  I supplied a diff that you can use to add some comments to the new tests if you like.16:46
adeuringbenji: thanks!16:46
rick_h_frankban: still around?16:47
frankbanrick_h_: yes16:50
rick_h_frankban: nvm, found it. Case mis-match in grep 16:50
frankbanok16:50
rick_h_jcastro: your bundles for jenkins and mediawiki still have the cpu= constrints16:55
jcastroack, I'll blow em away in a minute16:56
rick_h_jcastro: thanks16:56
frankbangary_poster: juju upgrade-charm + juju set juju-gui ga-key=UA-1018242-44 builtin-server=false run ok. to be noted that in the process the latest (local) release of the GUI was also installed16:59
hatchdoes anyone know of any broken charms or a way I can cause an error in a real env?16:59
rick_h_hatch: how so? We try not to ingest broken charms 16:59
rick_h_hatch: how are you looking to trigger an error?16:59
gary_posterhatch, mediawiki broke for me yesterday :-/16:59
hatchtrying mediawiki17:00
gary_posterfrankban, yay, thank you. so, looking at https://wiki.canonical.com/InformationInfrastructure/WebOps/CDO/JujuGui at bottom... sorry, on call, one sec...17:02
hatchI think we need to write some broken charms to test hook failures and the like17:04
gary_posterfrankban, I am going to edit the instructions and then ask you to approve diff17:04
frankbangary_poster: ack17:04
gary_posterthx17:05
hatchunfortunately mediawiki worked just fine :/ hah17:13
gary_postersorry :-)17:19
gary_posterhatch make a charm and break it?  use an old revision of the juju gui charm that doesn't support juju core?17:19
hatchI'd really like a testing charm which allows us to test failures17:20
hatchI have the string format of the StatusInfo17:20
hatchbut StatusData is implementers choice :)17:20
gary_posterfrankban, http://paste.ubuntu.com/6416799/ has ballpark side-by-side diff, which omits chars over 80.  http://paste.ubuntu.com/6416801/ is new.  http://paste.ubuntu.com/6416811/ is original.17:21
frankbangary_poster: looking17:21
gary_posterhatch, so make a testing charm! :-) or use an old revision of the charm you made that was broken at one point, or introduce a new error in a branch of it17:21
gary_posterI can push it up if you want :-)17:22
gary_posterthank you17:22
hatchpssht, I don't write broken anything17:25
hatch:P17:25
gary_poster:-)17:25
hatchyeah I'll write a broken one17:25
frankbangary_poster: the new instructions look good. the only thought I have is that currently they have juju-gui-source set to something (a url IIRC). maybe we want them to switch to local? 17:27
gary_posterfrankban, ah!  yes, thank you.17:28
gary_postergreat catch17:28
frankbangary_poster: in my test, I did not set juju-gui-source, so my test worked well with stable. I'd assume it should work also fi they switch on the fly from url: to local.17:29
gary_posterok thanks17:29
frankbannp17:30
bacfrankban: hey got a sec still?17:41
frankbanbac: on call, will ping you17:41
bacok, thanks17:41
=== rogpeppe1 is now known as rogpeppe
gary_postercharmworld update starting now-ish. eek17:56
rick_h_charmworld? or the jujucharms.com?17:58
* rick_h_ looks again17:58
bacwow, condesation from my cup fell on my MBP track pad and it went crazy with fantom events.18:01
bacs/condesation/condensation/18:01
jcastrorick_h_, oh bummer, in the jenkins bundle the constraints are blank18:03
jcastrodoes that still mess stuff up?18:03
rick_h_jcastro: that's ok, but the cpu one gets rejected18:04
rick_h_so you can't deploy it18:04
jcastrooh right!18:04
jcastroit's cpu-cores18:04
jcastroon it18:04
hatchevilnickveitch: can you pm me your email and I can get you the list which was made 2 weeks ago18:08
hatchthen I can expand on it later18:08
frankbanbac: ping18:12
bachi frankban18:12
bacfrankban: can you look at this real quick http://paste.ubuntu.com/6416921/18:13
frankbanbac: sure18:13
bacfrankban: i expected calls to increment_deployment_counter to return a boolean but instead i'm getting a tornado.concurrent.TracebackFuture18:14
frankbanbac: ok = yield utils.increment_deployment_counter(bundle_id, cw_url)18:16
frankbanbac: it seems yield is missing: each gen.coroutine returns a future18:17
bacfrankban: ah, right!18:18
frankbanbac: also you can use mock.Mock(return_value=BadResponse) instead of defining your own mock function. <shrug>, but then you can take advantage of assert_called_once_with18:19
evilnickveitchhatch, ok18:20
frankban(and IIRC FakeResponse can also be mock.Mock(code=404)) etc...18:21
gary_posterfrankban, jujugui, jujucharms has new stuff.  yay!18:22
rick_h_gary_poster: woot!18:23
frankbangreat!18:23
rick_h_gary_poster: now to send the email "STAY OFF COMINGSOON IN FRONT OF CUSTOMERS" :)18:23
rick_h_or large tech conference gatherings 18:23
frankbangary_poster: so from now on updating jc.com will also mean updating the GUI there. sounds cool18:24
gary_posterfrankban, you mean 'cause the tarball is there, yeah18:24
frankbangary_poster: yes18:24
frankbancool, time to go, bac: is everything ok?18:25
bacfrankban: yes!  i may ask someone here to review it but ask you to do a 2nd tomorrow.18:26
frankbanbac: ok18:26
gary_posterhatch, you have been volunteered to brainstorm about Le Web. :-) if you have any ideas, please follow up with that mail from Sally18:30
hatchon it18:30
gary_posterthanks18:31
rick_h_oh oh oh, it's working finally!18:40
frankbangary_poster: just noticed bundles don't work in jc.com: Object #<PythonEnvironment> has no method 'deployerImport' . we might want to ask for an apiBackend change in config js and/or applying a patch like this http://pastebin.ubuntu.com/6417214/ to the charm (not tested)18:45
gary_posterfrankban, ah frak :-/ thanks will pursue18:46
frankbangary_poster: ok thanks, have a nice evening18:48
gary_postertahnks frankban !18:48
gary_posteryou too!18:48
rick_h_jcastro: does the discourse charm take a while to setup in a live deploy?18:49
rick_h_jcastro: nvm, github goes boom and this charm goes with it :/18:52
jcastrorick_h_, yeah, about 20 minutes19:16
jcastrorick_h_, it doesn't do much until it relates to postgres19:16
jcastrothen it does a bunch of stuff19:16
rick_h_jcastro: yea, one of it's deps hit github which was 500'ing and it took a LONG time to timeout and die19:16
rick_h_jcastro: yea, just testing it as a bundle and making sure my in-progress/bundle complete notifcations are working19:16
jcastroyeah, well, it's rails, the entire platform depends on the right dice roll of servers being up19:16
rick_h_yea19:17
gary_posterjujugui, looking for quick charm review of https://codereview.appspot.com/26430044/19:19
baci will19:19
gary_posterthank you19:19
gary_posterhey benji.  ready any time.  sorry.  lemme know when you are ready19:20
bacgary_poster: quick enough?19:20
benjigary_poster: sure, one minute19:20
gary_posterbac :-) thank you yes19:20
gary_posterMakyo, approved swap19:21
Makyogary_poster, ty :)19:21
gary_posterit was a hard choice ;-)19:22
* gary_poster moving all releasable kanban cards to done...leankit is working hard...19:22
gary_posterheh, "An error occurred while cards moving."19:22
gary_posterOur massive card mountain proved too much for the poor little kanban board...19:23
benjiheh19:24
gary_posterjujugui, the releasable lane is clear :-) (much rejoicing etc.)19:26
rick_h_hatch: Makyo ok, I think I've fixed the final issue. QA instrucitons noted in the MP now https://codereview.appspot.com/26290043/ if you guys can start review19:37
rick_h_hatch: Makyo I'm running my hopefully finally fresh QA from scratch, but the code is near enough final and ready for full review and such. Only need one QA if hatch wants to bribe Makyo to do it for him :)19:38
bacbenji: i manually hit some URLs on staging yesterday to increment metrics and saw them work by getting the 'total'.  today they are back to zero.  is that expected?20:02
hatchrick_h_: ok I'll do the review first and see where that gets us20:03
bachatch: man, that mayor in toronto is getting classier by the day!20:03
hatchbac: hah, I don't really follow it but it sure sounds like he likes to party20:04
rick_h_bah, one more thing to fix20:05
rick_h_jcastro: so nope, can't have the empty constraint values. Thave have to be legit20:06
jcastrorick_h_, .... so basically, you need me to remove them entirely?20:15
jcastrorick_h_, is this a deployer bug?20:15
rick_h_jcastro: it's a mix I think. Yes, we should remove them as that'll update in 15min vs a production deploy20:16
rick_h_jcastro: the thing is that Go won't accept an empty value for the key20:16
jcastrook so you want me to do it now?20:16
rick_h_jcastro: so we should probably rip empty ones out along the path to juju-core before it gets them20:16
rick_h_jcastro: yes please, they're up in the store for people to see and they won't work20:17
rick_h_and we've been telling sales folks to use them to demo, but if they do it in a live env, it'll die20:17
jcastrowait, what?20:17
rick_h_right now, is someone were to bootstrap an ec2 env, and deploy your bundle, it'll fail silently20:18
jcastrodone!20:19
rick_h_jcastro: ty much kind sir20:19
bacrick_h_: did you say jcastro's cloudfoundry bundle was super slow to deploy?20:21
rick_h_bac: yea, takes about an hour to finish I think20:21
bacgrrr20:21
baci used it for a test since it only has one service20:22
jcastrobac, it takes an hour20:22
jcastroif you want a fast one do mediawiki20:22
bacjcastro: yours?20:22
jcastromediawiki-simple to be precise20:22
rick_h_bac: lol, looks can be deceptive20:22
jcastroyea20:22
hatchrick_h_: is putting each param on a new line some python thing? :P20:22
bacjcastro: mediawiki bundle?20:22
rick_h_hatch: :P it's a reads nice thing when you can't fit on one line to start 20:23
bacah, mediawiki-simple20:23
hatchhaha20:23
jcastrobac, if you have it locally please bzr pull, I just updated it about 2 minutes ago20:23
bacjcastro: i need to deploy from staging for my test20:24
benjigary_poster: ooh, one other thought I had in my notes: the "connect to a local juju environment from the GUI" feature doesn't have to be limited to the LXC provider, therefore we can let people manage their own EC2 (or private cloud) from a hosted GUI20:31
gary_postertrue, benji20:32
gary_posterthanks20:32
rick_h_gary_poster: heads up, filing this on the charm, but might be a deployer fix, not sure atm. https://bugs.launchpad.net/charms/+source/juju-gui/+bug/125142020:37
_mup_Bug #1251420: reporting an error from the environment isn't formatted well <juju-gui (Juju Charms Collection):Triaged> <https://launchpad.net/bugs/1251420>20:37
gary_posterack rick_h_ thx20:38
rick_h_gary_poster: going to move forward with my branch as is, it shows that poor error message to the user for now since it's not my branch's fault? Is that ok or would you prefer I hide the message for now?20:38
gary_posterrick_h_, go forward, thx.  what you see is <Env Error - Details:\n { u'Error': u'json: cannot unmarshal string into Go value of type uint64',\n u'RequestId': 1,\n u'Response': { }}\n > ?20:40
rick_h_gary_poster: yea, you see "The deployment errored: xxxxxxxxxxxx"20:40
rick_h_in the notification dropdown20:40
rick_h_jcastro's bundles working all my code paths in QA so well :)20:41
jcastrorick_h_, I am debating stealing maarten's and putting them in there20:41
gary_posterjcastro, I copied quickstart to https://launchpad.net/~juju/+archive/stable20:41
rick_h_jcastro: cool, gary has a working version of a couple of them I think20:41
jcastrobut I don't want to just shove them in there because20:41
jcastroI'd rather have "real" ones that people actually use20:41
jcastroinstead of "here's what I think big data should look like even I know nothing about big data."20:42
jcastrogary_poster, \o/20:42
gary_posteryeah jcastro I put some of Maarten's in "demo".  Can add more if you like.  https://jujucharms.com/sidebar/search/?text=demo20:42
jcastrorick_h_, wanna sync up on G+ wrt. quickstart? I'd like to write it up now that everything is landed20:42
rick_h_jcastro: sure thing20:42
gary_posterok quitting time.  ttyl20:43
rick_h_gary_poster: have fun20:43
gary_posterthanks you too20:43
=== gary_poster is now known as gary_poster|away
jcastrohttps://plus.google.com/hangouts/_/7ecpjr7uvbg386q48qdcdnf7qc?hl=en20:43
bacjcastro: on ec2 how long should it take to see some action when deploying your mediawiki-simple bundle?  i got nothing.20:55
jcastroit's about 7 minutes for the bootstrap20:56
jcastroand then after that they should come up quicklyish. 20:56
rick_h_bac: wait on ingest cycle, it's got a constraints bug in it atm20:57
bacjcastro: that's deploying from the gui?  i've hit the deploy button, it was acknowledged, but nothing20:57
bacrick_h_: should it give feedback?20:57
rick_h_bac: yea, there's empty constraints that it fails on20:57
rick_h_bac: welcome to my branch :)20:57
bacrick_h_: ok, good time to walk the dog20:57
bacrick_h_: but shouldn't it give feedback if it fails with constraint errors?  or does it just silently slink away?20:58
rick_h_bac: well, is this in sandbox or real env?20:59
bacec220:59
rick_h_bac: in a real env, it hits Go and Go up and dies. and my branch is working on displaying that Go error back to the user20:59
bacah, gotcha20:59
rick_h_bac: so right now, there's an error thrown, but the gui isn't listening for it and my branch is working on that atm20:59
bacrick_h_: i've also changed charmworld-url to point to staging.  lots of things to go wrong.21:00
rick_h_bac: he's updated his bundle so once it ingests the constraints should go away 21:00
rick_h_so that bundle should work shortly21:00
bacok great21:00
Makyojujugui trying to QA in an lxc, anyone have experience with this? lxc-start: command get_init_pid failed to receive response21:01
rick_h_Makyo: no, never seen it. I know quickstart doesn't support lxc if you're trying quickstart?21:02
Makyorick_h_, just trying to get an environment without juju so I can qa quickstart installing it.21:02
hatchrick_h_: ok almost done the review - am I missing why we can't process deltas for these updates instead of longpolling?21:12
rick_h_hatch: ok, heading out. I landed one small change to fix using the import button21:12
rick_h_hatch: we're not longpolling are we? I'm not sure how deltas work tbh. This was the api spelled out in the card from frankban21:12
hatchNote: This returns once the server has something to update on. It might wait a while.21:13
rick_h_hatch: see http://bazaar.launchpad.net/~juju-gui-charmers/charms/precise/juju-gui/trunk/view/head:/server/guiserver/bundles/__init__.py#L12721:13
rick_h_hatch: true21:13
rick_h_hatch: so no idea, this is what I had for instructions to work from21:13
hatchok maybe I'll let this sit for tonight and ask him in the morning21:14
rick_h_hatch: I dont' think there'sa way in the delta to tell is a change was from a bundle request or just a command21:14
rick_h_hatch: so just watching the delta as it exists will do me much good tbh21:14
rick_h_hatch: especially since you can run multiple bundle deploys at once and have several running around at once until they complete21:14
hatchright but can't we send the data down the delta that's being returned from this rpc call21:15
hatchI don't know, I'm curious as to why that's not the case21:15
rick_h_hatch: probably because this is run by the guiserver and the delta is proxied through it, but not part of it21:15
rick_h_hatch: so the guiserver could mix up requestid and such if it tried to use the same channel21:15
hatchright so we should be able to send a fake delta21:16
rick_h_hatch: possibly I guess, I'm not sure if that's feasible from the guiserver/deployer side to generate fake deltas that look legit/work from the go env side21:16
hatchI'm sure there is a valid reason why we aren't doing that but curious as to what that is21:16
* rick_h_ doesn't know enough about the delta stream21:16
hatchwould simplify this :)21:16
hatchcrap half my notes are on one rev and half are on the other21:17
rick_h_hatch: :P well now you tell me on day 3 of working on it 21:17
hatchyou'll have to decode them :P21:17
rick_h_hatch: all good21:17
rick_h_hatch: hopefully not that many notes :P21:17
hatchhaha no - I am probably goign to lgtm it21:17
hatchi havent' decided yet21:17
hatchlol21:17
rick_h_heh, well feel free to hold off. I'm going to EOD and do another round of QA in the morning. Kept finding 'one more small thing' each fresh deploy today21:18
rick_h_and still need the second review and a 3rd party QA so won't go anywhere until later tomorrow21:18
hatchsure thing I'll get the review finished for sure21:18
hatchbut not sure about qa today21:18
rick_h_rgr, all godod21:18
hatchI'm probably going to go back to lying down21:18
rick_h_all good21:18
rick_h_just mean no hurry21:18
rick_h_yea, cool. Go medicate and such. I'm out of here. Will push this boulder farther up the hill tomorrow21:19
bacrick_h_: this time it popped right up21:38
rick_h_bac: cool21:42

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