=== thumper-gym is now known as thumper === CyberJacob|Away is now known as CyberJacob === mwhudson is now known as zz_mwhudson === CyberJacob is now known as CyberJacob|Away === nate_finch is now known as natefinch [14:50] Does the config option need to be defined in a yaml file before it can be settable? I'm charming a service with nested configuration options. I have a "channels" config option which is a coma-delimited list encapsulated in a string but for each channel I would like to have a "_devices" config option which encapsulated a coma delimited list inside a string [14:51] timrc: yes [14:52] everything should be in config.yaml so you can't have pseudo-lists [14:53] mgz, Can I specify multiple config yamls? (e.g. config.yaml has the base config and then a local.yaml which not part of the charm but is used to deploy a specific instance/set of instances) [14:53] er as the base config* [14:53] no. [14:54] you can just be cleverer about what you're packing into your string [14:54] mgz, ugh [14:55] there might be a neater way of doing what you're attempting, that some experienced charmer could explain if you give the whole problem [14:59] mgz, I'm just going to use a ConfigParser-type config file, point to it in the config.yaml, and have my config-changed script read and Do The Right Thing (tm) [15:00] timrc: config.yaml just describes the config [15:00] your charm needs to get the values out when hooks run using the juju commands [15:16] marcoceppi: im' getting a key error on admin-secret, i'm in the local environment for juju, and the trace is: http://paste.ubuntu.com/6750920/ [15:17] this is regarding an amulet test - btw. not juju core, i should specify that. [15:17] lazypower: yeah, this is a deployer error, do you have an admin-secret key set for your local environment? [15:18] i dont want to answer that... [15:18] i do not (facepalm) [15:19] admin-secret is used by deployer because it's the password for gui [15:19] that's why deployer is failing, since amulet uses deployer to drive deployments you have your answer [15:19] * marcoceppi makes note to make deployer fails nicer [15:19] That i knew. My environments.yaml has changed from what i remember it being, or I'm thnking of my juju lab. 1 sec while i run this down. [15:23] man, what happened to the queue [15:23] we were doing so well! [15:27] jcastro: all in due time my friend [16:09] marcoceppi, jcastro: How do I bootstrap juju on trusty. I know how to make a charm go to trusty, but not bootstrap the state-server on trusty [16:14] the bootstrap node is trusty [16:17] man I think i just hozed my juju installation :| I've got a leftover in here somewhere after downgrading from 1.17 thats breaking my local provider [16:17] ERROR TLS handshake failed: x509: certificate signed by unknown authority [16:21] lazypower: delete ~/.juju/environments/local.jenv [16:21] sinzui: default-series: trust? [16:22] I've never tried to make a trusty bootstrap node [16:25] marcoceppi, yeah, that is the only way I know too [16:25] sinzui: did it not work? [16:26] marcoceppi, we haven't checked yet. actually, I do know that method works. We wanted to use an existing env in yaml, but just change the series [17:25] sinzui, does bug 1254401 only affect 1.17? [17:25] <_mup_> Bug #1254401: error reading from streams.canonical.com

[17:25] sinzui, specifically is 1.16 not effected [17:26] arosales, it affects 1.16+ it will be fixed when we deploy to the site [17:27] sinzui, so all deploys are dead in the water this whole week? [17:27] arosales, there is nothing for a developer to do the bug cannot be in progress, or if it is, utlemming is working on it [17:27] arosales, the site has never been up [17:27] I am just wondering if folks know the severity [17:27] arosales, the bug was filed months ago [17:27] sinzui, understood but 1.16 worked before . . [17:28] arosales, juju always tries the tools-url, then streams.canonical.com then aws [17:28] oh, it looks in the control bucket first [17:29] sinzui, oh so deploys should still work after streams.c.c fails [17:30] sinzui, sorry I thought deploys were broken all together [17:30] arosales, CI has never failed because the site doesn't exist [17:30] sinzui, whew [17:30] sinzui, I understand now, thanks for clearing that up. [17:30] arosales, log show a a lot of hate, then juju comes to terms with the issue and moves on [17:31] sinzui, gotcha and thanks for the education [17:32] mbruzek, note your hp deploy should succeed after the errors ^ [17:33] right according to the workaround in the bug 1254401 I was able to get this to work with the following [17:33] <_mup_> Bug #1254401: error reading from streams.canonical.com

[17:33] juju bootstrap -e hpcloud --upload-tools [17:37] mbruzek, no need to that [17:37] no? [17:38] juju bootstrap -e hpcloud ended in Error the first time I tried it [17:38] juju sync-tools didn't work either. Same error [17:39] mbarnett, each CPC has a local copy of tools. Looks like Juju doesn't check for the CPC tools, but you can tell it to [17:39] mbruzek, ^ add tools-url: https://region-a.geo-1.objects.hpcloudsvc.com/v1/60502529753910/juju-dist/tools [17:39] I was only able to get it to work with the --upload-tools [17:39] mbruzek, I understand that feature will be removed in the future, but will work for a few more months. it is often abused [17:40] OK [17:40] mbruzek, tools-url will be renamed tools-metadata-url. It will always to the right thing and do it quickly [17:41] OK so add tools-metadata-url: https://... to my environments.yaml file under hpcloud? Or tools-url? This is for trusty and 1.17? [17:44] adding tools-url [18:00] mbruzek, hey so devel release questions will be closed on askubuntu, things like that where stuff is fast moving should be on irc or the mailing list. [18:01] I was trying to document this for others, and build rep. [18:02] yeah that won't work. :) [18:08] jcastro: should this be removed then? http://askubuntu.com/questions/405472/how-can-i-downgrade-my-juju-revision [18:08] I would drop the specific versions [18:08] and make it more generic [18:08] "How do I revert from a development version of Juju to a production one?" [18:10] Thanks. Completed. [18:53] hey, juju-core guys, what does JUJU_HOME expect as a value? With or without the .juju directory? natefinch? === CyberJacob|Away is now known as CyberJacob [18:55] marcoceppi: with [18:55] natefinch: ta! [18:55] marcoceppi: /home/foo/.juju is the default value for JUJU_HOME [18:56] natefinch: that's what I was looking for, awesome [18:56] i.e., if you set it, we don't put a .juju folder in the folder we specify, we just dump data directly in that directory [18:57] s/we specify/you specify [19:40] mbruzek: I couldn't get hpcloud to bootstrap with tools-url, were you able to get a bootstrap? [19:40] Yes only with the --upload-tools [19:41] It was already running from that, so my tools-url is untested [19:42] mbruzek: nvm, I got it working had to remove the .jenv file [19:43] marcoceppi, the postgres review wasn't as heavy as I thought it would be [19:43] so I just pushed [19:43] * jcastro cowboys [19:44] jcastro: yeah, stub does a great job maintaining the charm already [19:51] _thumper_: did the default behavior of destroy-environment change from using a -e flag to just a parameter? === _thumper_ is now known as thumper [19:51] I could have sworn it used to be -e flag [19:51] marcoceppi: yes [19:51] marcoceppi: the name of the environment is now always needede [19:51] thumper: y u do dis [19:51] marcoceppi: no, nate [19:52] marcoceppi: primary reason was to make someone type 'juju destroy-environment production' [19:52] and just like that, half of my scripts broke because -e is no longer a valid flag [19:52] instead of just getting the name from the environment [19:52] https://juju.ubuntu.com/docs/charms-destroy.html#destroy-environments needs updating too [19:53] thumper: when did this change land? or is it a 1.17 thing? [19:53] I can't recall this being an issue in 1.16 [19:53] marcoceppi: I think so [19:54] ohhh, this will be tricky [19:54] can I file a regression that -e should be valid with no parameters passed for env? [19:54] thumper: as in, do you think that would be addressed by next stable? [19:55] or, is it even possible to do [19:55] anything is possible, just not simple [19:55] is it simple* to do? [19:55] no [19:55] well... kinda === zz_mwhudson is now known as mwhudson [19:56] marcoceppi: feel free to file the bug, and we can discuss it [19:56] thumper: Okay, I'll file it as a bug then and hope it gets prioritized for a 1.18 cut [19:56] thumper: thanks o/ [19:56] np [20:01] <_mup_> Bug #1269119 was filed: destroy-environment no longer accepts an -e flag [20:02] damnit, wrong project [20:17] maxcan: you around? [20:18] I am [20:18] whats up? [20:18] you're question has the horrifying implication that I'm considered someone who knows something about juju [20:21] * lazypower grins [20:22] You're doing docker + juju "loose integration" in your production stack arent you? [20:22] what do you mean by "loose" [20:23] we have some charms that where we use a binary that we "ship" as docker images [20:23] the charm downloads and runs that docker image [20:23] if thats what you mean [20:23] i've thought about writing a generic docker image charm [20:24] but it wouldn't work for us since we require authnetication to d.load the charm etc [20:24] Writing an amulet deploy test, I got an error when trying to create a relation [20:24] ValueError: Can not relate, service not deployed yet [20:25] mbruzek: can you pastebin the test? [20:25] maxcan: ah ok. I thought i saw environment exports for the config and relation changes [20:26] mind you that was what, a week ago at 11pm EST :) [20:26] http://pastebin.ubuntu.com/6752404/ [20:26] marcoceppi, ^ [20:26] mbruzek: you reference rabbitmq but you deploy rabbitmq-server [20:27] lazypower: you mean on the MMS charm? [20:27] Thanks marcoceppi [20:27] mbruzek: either change the add line to d.add('rabbitmq-server', 'rabbitmq') or change the d.relate to use the rabbitmq-server name [20:27] lazypower: oh, for our main charm? yes, we export env vars into hte docker container with -e [20:27] If I had changed the name on the deploy that would have worked [20:27] s/main/docmunch-yesod [20:28] mbruzek: actually, it's d.add('rabbitmq', charm='rabbitmq-server') [20:28] sorry about that [20:28] * marcoceppi needs to read his own docs [20:31] lazypower: i was about to step out for lunch, anything i can help with before I go? [20:32] That was it my man. I had a cursory question about your docker usage with juju [20:32] Thanks! [20:32] np [20:43] thumper: the output a user gets during bootstrap,that's you, right? [20:43] marcoceppi: as in, did I do that? [20:43] no, that was axw [20:43] yeah [20:43] aw, next time I see him, let him know it's really nice [20:43] don't have to scratch my head anymore [20:44] \o/ [21:03] marcoceppi: hey, did you get a chance to review that storage charm? :) [21:06] dpb1: he did, he's pending a paper writeup for you. [21:06] oh no [21:07] lazypower: I responded to your review comments on the swap charm, thx. :) [21:07] and its not "oh no" :) you're doing some interesting new stuff in there [21:08] lazypower: yes, agreed. We are eagerly anticipating the results. [21:10] Thank you for submitting the swap fixes so quickly! [21:10] I'll try to get that re-reviewed by tomorrow morning, bit backed up atm [21:11] lazypower: no worries, that one is pretty simple. [21:23] OK I have a new amulet test written ( http://pastebin.ubuntu.com/6752656/ )with the relations corrected. When I run the test I see it getting started on the hpcloud but I run into this error. [21:23] AttributeError: 'module' object has no attribute 'CalledProcess' [21:24] mbruzek: can you provide the full trace? [21:25] http://pastebin.ubuntu.com/6752675/ [21:25] my admin-secret is not setup correctly? [21:27] We need that for the local one, but I set this environment up for hpcloud where juju uses the keypair [21:27] So set something for admin-secret and do I need to rebootstrap? [21:27] mbruzek: all environment s need an admin secret [21:28] yes [21:28] OK === CyberJacob is now known as CyberJacob|Away === CyberJacob|Away is now known as CyberJacob === mwhudson is now known as zz_mwhudson