=== thumper-bjj is now known as thumper [10:11] jamespage, would you agree that the charms pep8 tests shouldn't cover charmhelpers? [10:11] gnuoy, yes its normally excluded [10:11] see tox.ini [10:47] beisner, I don't understand my -1 from https://review.openstack.org/#/c/289229/ . Is that Openstack CI failing to initiate tests remotely? [10:57] jamespage, beisner, I want to run a mojo spec using my branch which is up for review at review.openstack.org. I'm following the workflow where the only remote location of the wip branch is whereever gerrit stores it when the mp is created. So, I need to tell mojo where to check out my wip branch from. I see no reference to my wip branch on gerrit. Any ideas? [10:57] gnuoy, one sec yes [11:02] gnuoy, ok [11:02] so for you current review: [11:02] git clone https://review.openstack.org/openstack/charm-keystone [11:02] cd charm-keystone [11:03] git fetch https://review.openstack.org/openstack/charm-keystone refs/changes/29/289229/2 [11:03] git checkout FETCH_HEAD [11:04] jamespage, thanks but I don't see how that is going to work with mojo [11:05] gnuoy, if you want a straigh branch to ref, the maybe just fork on github and push the branch to the fork for wip testing [11:05] ack [11:06] gnuoy, that's what I'm doing for https://github.com/javacruft/charm-ceph/tree/network-spaces [11:06] for example [11:06] want to smoke it before staring to consume ci cycles for testing [11:07] gnuoy, do you know had to add additional remotes to your local repo? [11:07] yep [11:08] good-oh [11:28] jamespage, that's working, thanks [12:19] dosaboy, thedac, gnuoy: http://paste.ubuntu.com/15320377/ well that works... [12:19] maas-spaces2 branch + MAAS 1.9 [12:22] jamespage: congratz [12:22] dosaboy, containers support is just around the corner, and then we can flesh out a openstack-base bundle [12:57] thedac, gnuoy: https://code.launchpad.net/~james-page/charm-helpers/network-spaces/+merge/288285 [12:58] that's what we're getting for Juju 2.0 re network-spaces... [14:11] nagyz: the MAAS CLI drives an API directly. You could script something directly against that if you were looking to simplify/automate something [14:28] hi gnuoy, re: that initial -1, looks like it was an interwebs glitch http://pastebin.ubuntu.com/15321038/ [15:01] jamespage -its simple. I like it [15:02] lazyPower, ? [15:02] oh network-spaces [15:02] the network-spaces mp you linked above [15:02] yes [15:02] I agree it is simple [15:02] so instead of relation-get, we network-get and it hands back the IP address it should bind to? [15:03] jamespage, our mojo specs define the neutron-gateway:mysql shared-db relation. that relation found retirement last wk, so we'll need to update all next topologies in spec tests, yah? [15:03] o-c-t as well, if it's there [15:04] beisner, oh I fixed o-c-t already [15:04] needs dropping everywhere.... [15:04] lazyPower, yeah I think so [15:05] jamespage, ok cool, thx for the o-c-t updates. i'll work on the mojo spec updates. [15:06] lazyPower, you always need the name... [15:12] lazyPower, " jamespage, looking [15:12] jamespage, LGTM" [15:12] lazyPower, wanna give me a charmers +1 so I can land that? [15:12] thought i'd missed the deadline for a talk submission then realised March comes before April..... [15:13] sure, where does teh binding param come from? is it just the name of the relation? [15:14] lazyPower, it can be a relation or extra-binding name [15:15] lazyPower, relation or 'extra-binding' which is a relation-less network binding mechnism [15:15] ha [15:15] sorry - dobuled up where [15:15] there [15:15] can't type... [15:15] It happens to me all the time :D i'm contageous [15:17] dimitern, I really really want to play with the containers support now :-) [15:18] jamespage, I know :) as soon as that's in a testable state will let you know - should be within the next week or so [15:19] jamespage - +1'd [15:19] lazyPower, lovely thanks... [15:34] i want to test leader election in amulet [15:34] is there a way of asking juju which unit is the leader? [15:35] magicaltrout - `is-leader` [15:36] er [15:38] magicaltrout: not easily, at least easily atm [15:38] thats fine i set it in a status message anyway [15:38] so I'll grep the status [15:38] just wondered if there was an official way [15:39] juju doesn't really expose who the leader is externally, so we could make a d.service['service'].leader property, or something similar, where we abstract the testing of that away [15:39] magicaltrout: I'll open a bug to track [15:40] yeah i looked at juju status yaml and didn't see anything of note [15:41] unless you status-set something of value wrt leadership, there is none [15:41] yeah thats what i do [15:41] jam's LXD patch is f***ing rocking. Back to LXD Local \o/ [15:42] heck yeah! happy to hear that [15:43] it should squeak into beta2 due out this weekish [15:44] * magicaltrout lives in the land of adhoc patching and trunk builds, beta's have no meaning on this server....... [15:48] hah [15:52] magicaltrout: bonus if its production. [15:53] jrwren: hehe, no, i'm not that crazy........... [15:53] my prod server is 2.0 Alpha1.....! ;) [16:08] marcoceppi: i'm clearly just not grepping amulet properly [16:08] i know about wait for status [16:08] how do i just get the status? [16:09] or the message, to be more precise [16:12] magicaltrout: I don't remember, let me check [16:12] magicaltrout: ['workload-status'].get('message') [16:13] thanks, i was just in the process of getting amulet to print me the contents of the unit object [16:14] magicaltrout: basically, unit is the contents of the yaml output of juju status for that unit [16:14] so workload-status and message dict key from that [16:28] marcoceppi: hi, trying to figure out the syntax for the placement stuff from what you suggested. Any pointers as to constraint to use for machine name? [16:28] larrymi: I'm not sure, that used to be a secret constraint in juju but it may have been removed. If you use tags in MAAS though it'd have the same effect [16:33] marcoceppi: I see thanks. I tried the tags for 5 machines but getting "ERROR cannot deploy bundle: machine "0" is not referred to by a placement directive (and 4 more errors)" [16:49] tvansteenburgh: There's several charms showing up in the review queue but have been merged: 2433, 2430, 2432, 2299 [16:53] https://gist.github.com/buggtb/ec2b9d5abb77c058b46e marcoceppi unleash your python-fu and spot the mistake please :) [16:54] well, i get the error, but I can't figure out the correct syntax :) [16:56] line 2 is missing a closing ) [16:56] magicaltrout: https://gist.github.com/marcoceppi/298d7db8b409900b2a99 [16:56] aisrael: i had to type it, ignore the typos the code compiles :) [16:56] also, I think this: self.d.sentry.unit['pdi/0'] should be: self.d.sentry.unit['pdi'][0], but don't hold me to that [16:57] aisrael: ^ ;) [16:57] booo [16:57] .unit is pretty outdated at this point. self.d.sentry['service'][#] is better [16:57] the amount of different variations i tried [16:57] I dint use .unit initially but they both returned the same error anyway :) [16:59] magicaltrout: yeah, .info isn't really that obvious, but because self.d.sentry['service'][#] is a UnitSentry class, it wasn't originally made to be subscriptable, https://github.com/juju/amulet/blob/master/amulet/sentry.py#L63 so we placed the data in .info - that's a design decision from 3 years ago. Obiviously self.d.sentry['service'][#]['workload-status'] should just work [17:00] yeah i would have taken about 3 hours to guess that I reckon [17:01] works though... thanks a lot! [17:03] aisrael: updated those reviews, thanks [17:03] magicaltrout: i am working on api docs for amulet as we speak - hopefully that will help [17:40] hi thedac, can you have a look over this mojo spec review?: https://code.launchpad.net/~1chb1n/openstack-mojo-specs/mojo-openstack-specs-juju-wait-for-workload-status/+merge/288302 [17:40] beisner: sure [17:43] beisner: the resyncs from gerrit to lp, are they at next still or somewhere else? [17:44] thedac, @ next [17:44] thanks [17:44] ditto :-) [17:48] charmers: anyone willing to take a look at a quick bugfix? :) https://code.launchpad.net/~xfactor973/charm-helpers/ceph-configkey-bugfix/+merge/288146 [17:48] 8 liner :D [17:54] cholcombe +1'd [17:54] lazyPower, thanks! [17:55] beisner: so, ceph-radosgw has not synced since Mar 2. Can you take a look? [17:57] thedac, let me make sure we're on the same page(s). it looks in sync to me: [17:57] https://github.com/openstack/charm-ceph-radosgw/commits/master [17:57] to: https://code.launchpad.net/~openstack-charmers/charms/trusty/ceph-radosgw/next [17:58] Let me double check but the bzr log in the next version has a last commit of Mar 2 [17:58] thanks tvansteenburgh looking forward to them! :) [17:59] beisner: ignore me. I had reverse on the bzr log ;) [18:00] aisrael: I've got an MP for charmhelpers that adds resource-get [18:00] thedac, woot :) [18:04] aisrael charmers https://code.launchpad.net/~marcoceppi/charm-helpers/resource-get/+merge/288328 [18:09] beisner, thedac: I have a glance and ceilometer review in the queue which fix from bits of mitaka support - any chance you can review? [18:09] they both full smoke OK [18:09] o/ jamespage - sure [18:10] jamespage: yes, I am having issues logging into review.openstack.org. Trying to figure out what I am missing [18:10] thedac, hmm [18:10] sso matching maybe [18:10] Yeah, I get into a loop with sso [18:12] fyi jamespage, thedac, gnuoy - added [uosci-crit] failure email nag on the github-to-launchpad sync job [18:14] beisner: jamespage: is there an LP team we need to be in to log into gerrit? [18:14] I notice sso is not passing any teams [18:15] thedac, not afaik [18:21] beisner, here's the debug log: https://gist.github.com/cholcombe973/11f017456581397664f4 :) [18:21] beisner, rolling upgrade in action [18:21] it's noisy [18:56] marcoceppi: awesome, I'll check it shortly === alexisb is now known as alexisb-afk === alexisb-afk is now known as alexisb [20:33] is there any documentation we can look at to add MAAS to juju 2 ? all the docs we find and even the cli help is old [20:35] metsuke_ - there was a section about this on the release notes, one sec while i fetch you a link [20:35] metsuke_ - https://lists.ubuntu.com/archives/juju/2016-February/006618.html under private clouds [20:39] lazyPower: thanks [20:39] will juju 2 have multi-user authentication and management? [20:40] metsuke_: yes [20:42] rick_h__: cool, thanks! I saw the last on-air and the subject was kind of touched on. Specifically, can we link users in juju to openstack tenants or users? [20:42] metsuke_: ah not so much there [20:42] metsuke_: that the admin that bootstraps a controller can share/add other users to that controller [20:45] ok, how about models in juju lining up with openstack tenants? if that's not a think then how about imposing quotas on models to act like quotas on nova tenants? [20:46] metsuke_: so when you create a model you can specify cloud credentials to use [20:46] metsuke_: and those could be different tenants on OS and you could do things like enforce OS quotas I think [20:47] oh, so it is per model that we can specify the credentials. That would work just fine! [20:47] yea, it'll have a sane defautl, but be tweakable [20:49] Do you happen to know if there is a way to share volumes in juju or is that left up to your back-end (like Openstack Manila)? [20:51] so there's a roadmap feature for adding shared filessystems to the storage modeling Juju does [20:51] but yes, at the moment that's not available in juju itself [20:52] nice! In that case, I'll have to find someone to make a Manila charm [20:54] marcoceppi: I see a new charm on the horizon http://blogs.microsoft.com/blog/2016/03/07/announcing-sql-server-on-linux/ [20:54] rick_h__: we already have sql-server for windows ;) [20:54] sql server, the platform that makes grown men cry [20:54] marcoceppi: true enough [20:57] will the auto credential detection still require us to manually create a credentials file for MAAS? [20:59] stokachu: it looks at local files. wallyworld is there a list of what's supported? not seeing a list of paths in the release notes [21:00] i saw one for checking .novarc for openstack, so curious if there is one for maas === _thumper_ is now known as thumper [21:01] stokachu: don't see it but if there's a path please let us know and we can look to add https://github.com/juju/juju/blob/af8944dd5a0d6cfe65a1f0e6bd5116a479270a91/cmd/juju/cloud/detectcredentials.go [21:01] hmm that may need a feature request for maas guys [21:01] rick_h__: the release notes do list what's supported [21:02] i updated them last week [21:02] wallyworld: in the google doc? [21:02] yep [21:02] aws, google, openstack/rackspace [21:03] can i get access to that google doc? [21:03] just to read it [21:03] sure [21:09] re charm promulgate, it's throwing "ValueError: Series can not be None" when I run it inside the charmdir I want to promulgate. Is there more to the command? [21:09] marcoceppi: lazyPower ^ [21:10] aisrael charm promulgate -s trusty [21:10] lazyPower: mucho gracias [21:11] rick_h__: do the release notes answer your question? did you see the material? [21:12] wallyworld: looking at https://docs.google.com/document/d/1ID-r22-UIjl00UY_URXQo_vJNdRPqmSNv7vP8HI_E5U/edit under "autoload-credentials" it'd be good to list what clouds/paths it looks for [21:13] rick_h__: it does in the Credentials Management section [21:13] wallyworld: maybe a reference from that section to the other? [21:13] ok [21:13] ty [21:15] rick_h__: the info is a few lines above [21:15] ie the autoload feature is described in the credentials management section [21:15] where it looks is described just before the autoload command is talked about [21:16] there's no need for a link then is there? [21:16] ie all the info is together [21:22] cmars, menn0 http://reviews.vapour.ws/r/4085/ [21:22] mattyw: looking [21:23] menn0, thanks, I *think* it's what you suggested [21:25] mattyw: it is, but I've suggested further changes on the review [21:27] menn0, we still want to keep the DocID filed right? because we want to choose which field mongo wants to have as _id (our static key) [21:28] mattyw: when the document struct doesn't have an _id field, the "Id" on the txn.Op that inserts it is used [21:29] mattyw: given that you don't care about the _id when you read the doc out (it's fixed after all), there's no reason to have it on the struct. [21:29] mattyw: leaving it off the struct just avoids some clutter [21:31] menn0, I see it now, thanks for pointing it out [21:31] menn0, I'm going to have a pr with many more deletions than additions :) [21:31] mattyw: that's often a very satisfying kind of PR :) [21:32] menn0, just pushing the changes [21:36] menn0, PTAL http://reviews.vapour.ws/r/4085/ [21:44] wallyworld: hmm, ok yea it reads backwards to my brain but understand. I did a "find on page" for autoload and read down and so I never saw the stuff above it. [21:52] guys can you point me to a charm where I can learn (in python) how to change the variables for a given config template file ? or point me to a commit where a parameter was added to an existent charm ? [22:10] lborda: https://github.com/OSBI/layer-pdi/blob/master/reactive/pdi.py#L124 or something more? [22:14] humm... magicaltrout you're using the render function directly... cool how does that work if you have to juju service set var=bla ? [22:16] so if you look around line 58 [22:16] 61 etc [22:16] there are a bunch of data_changed hooks [22:16] to reprocess stuff if you change a charm setting [22:16] so I'd just use on of them and the render call [22:16] (I believe, i'm no expert but i'm pretty sure that would work fine) [22:17] the render calls down the bottom are triggered by leader election [22:17] which is pretty similar [22:21] magicaltrout, humm... lines 61..64 look like what I am looking for... :) thanks... it's actually for the keystone charm so I will probably have to use a variance of the logic ... but it helps a lot tks [22:24] no probs [22:24] magicaltrout, using charms.reactive is much easier... [22:26] hehe [22:26] well i got a very helpful double team training session from kevin and cory_fu in Ghent which helped a lot with picking the right stuff [22:26] especially as I knew no python [22:27] the only reason we wrote that layer in python is because i was stubborn and wanted to learn some python ;) [22:27] but its serving as a good place for me to refer to as its got a bit of everything [22:29] magicaltrout, doing it for learning python is a good excuse! :) tks === thumper is now known as thumper-dogwalk [23:08] evening all [23:08] o/ [23:08] is there a way to tell a test to wait until the next leader election routine has run? [23:09] or do I just tell it to chill out for 5 minutes and check again [23:18] Hm. You could watch the status for the next leader election to run, and then wait for that to finish