16:03:56 <b3rnard0> #startmeeting OpenStack Ansible Meeting
16:03:57 <openstack> Meeting started Thu Apr 23 16:03:56 2015 UTC and is due to finish in 60 minutes.  The chair is b3rnard0. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:03:58 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:04:01 <openstack> The meeting name has been set to 'openstack_ansible_meeting'
16:04:18 <b3rnard0> #chair cloudnull
16:04:19 <openstack> Current chairs: b3rnard0 cloudnull
16:04:31 <b3rnard0> #topic Agenda & rollcall
16:04:38 <palendae> hi
16:04:39 <b3rnard0> #link https://wiki.openstack.org/wiki/Meetings/openstack-ansible#Agenda_for_next_meeting
16:04:43 <cloudnull> o/
16:04:47 <Sam-I-Am> howdy
16:05:38 <sigmavirus24_awa> o/
16:05:42 <mattt> howdy
16:06:08 <d34dh0r53> o/
16:06:35 <b3rnard0> presente
16:07:27 <b3rnard0> we have no action items from last week. do we want to dive into blueprints?
16:07:48 <cloudnull> lets wait a min more.
16:07:55 <cloudnull> for people to join in
16:09:14 <Sam-I-Am> i think people melted
16:09:35 <cloudnull> ok. so lets go.
16:09:50 <b3rnard0> #topic Blueprints
16:10:09 <cloudnull> https://blueprints.launchpad.net/openstack-ansible/+spec/build-facts-archive
16:10:20 <cloudnull> ^ is jwagner around ?
16:10:23 <jwagner> here
16:10:36 <b3rnard0> #link https://blueprints.launchpad.net/openstack-ansible/+spec/build-facts-archive
16:10:51 <jwagner> never doen one of these meeting so...
16:10:55 <jwagner> do i just explain it?
16:10:58 <jwagner> done*
16:11:06 <palendae> I think that's a valuable addition. I would like to see maybe a little more detail about where that info is stored
16:11:46 <b3rnard0> jwagner: walk us through it briefly
16:12:26 <jwagner> i was imagining on the deployment host under /opt or something in a dir stucture of /opt/osad-archive/{hostname}/[archived-bits]
16:12:32 <jwagner> then tar it up
16:12:34 <jwagner> etc
16:12:47 <palendae> jwagner: So a set of yaml files?
16:12:58 <palendae> Or a text dump of facter's info?
16:13:42 <d34dh0r53> I think most of this is available via ansible facts, which can be run once and you'll get a report of all hosts in the inventory.
16:14:07 <jwagner> d34dh0r53 yes that is probably how we could get most of this info
16:14:32 <palendae> I know there have been requests to get this information into a database for deployer support personnel
16:14:37 <jwagner> my man concern and need was the ability to parse all this info and grab bits and pieces that we need to report bugs in a more readable and accessable fashion to you guys
16:14:48 <jwagner> also to be able to aggregate the tempest test results up to jenkins
16:15:25 <Sam-I-Am> when you say things like 'cinder config' do you mean the host-level stuff, the openstack config, or the actual volumes/usage/etc ?
16:15:37 <jwagner> currently I cannot get to the results due to the fact that our ansible plays has no info about the inventory so getting to the utility container that has the xml from the tempest run is pretty janky
16:15:39 <cloudnull> jwagner can you reformat/submit the spec this the spec repo using this template https://github.com/stackforge/os-ansible-deployment-specs/blob/master/specs/template.rst
16:15:51 <Sam-I-Am> some of this stuff is facts, but some of it is the result of slurping config files or API calls
16:16:28 <jwagner> cloudnull can do
16:17:16 <cloudnull> tyvm
16:18:08 <b3rnard0> #action jwagner -- can you reformat/submit the spec this the spec repo using this template https://github.com/stackforge/os-ansible-deployment-specs/blob/master/specs/template.rst
16:19:55 <b3rnard0> okay, any other blueprints for discussion?
16:20:17 <sigmavirus24_awa> palendae's blueprint?
16:20:20 <alextrcitiy> Can you guys check out the ceilometer bp spec when you get the chance?
16:20:25 <alextrcitiy> It's not approved yet
16:20:45 <cloudnull> #link https://review.openstack.org/#/c/169417/
16:21:10 <cloudnull> alextrcitiy my only issue with that spec is that we've not defined how testing will be done with mongo.
16:21:47 <palendae> I know we touched on this in past meetings
16:22:02 <palendae> Something like we were going to install Mongo as part of the gating?
16:22:09 <cloudnull> i think that if we deploy mongo via a gate script for the test case im happy. if we bring mongo as a role in im dubious.
16:22:16 <alextrcitiy> cloudnull: That's right. Are you looking for more description in the testing section?
16:22:21 <cloudnull> i am
16:22:31 <alextrcitiy> Okay sounds good
16:22:33 <palendae> cloudnull: Do you mean bring in an ansible galaxy role to do mongo gating?
16:22:36 <palendae> Or via bash?
16:22:49 <palendae> Or do you not care as long as it's not a role that gets installed everywhere
16:23:05 <cloudnull> either or, i just dont want the external role to be brought into the system.
16:23:24 <cloudnull> which will only create confusion when people look to deploy ceilometer.
16:23:27 <palendae> Right
16:24:10 <palendae> So, are you against including any galaxy roles in our install then?
16:24:13 <cloudnull> if we do bring in mongo as a role we support then it needs to be scalable.
16:24:30 <palendae> If so, we need to look at modifying the bootstrap ansible role to take that out
16:24:40 <palendae> And move it to a deployer-level solution
16:24:55 <palendae> bootstrap ansible script*
16:25:42 <cloudnull> +1
16:26:09 <cloudnull> but in terms of the spec for ceilometer the only thing im concerned with is how is it tested.
16:26:14 <palendae> Ok
16:26:19 <alextrcitiy> Gotcha
16:26:53 <cloudnull> so if we can get a test case into the spec im +2
16:27:14 <andymccr> you "can" deploy ceilometer with mysql though right?
16:27:24 <andymccr> i mean i know its not great in prod but for testing purposes/gating etc?
16:27:48 <cloudnull> andymccr you "can" but it wont work in horizon .
16:27:58 <andymccr> oh ok so no go even just for testing? :(
16:27:59 <mattt> andymccr: guess we'd need to test the whole mariadb installation mechanism in the gate to ensure that code path works
16:28:03 <palendae> Are we testing it in horizon?
16:28:05 <mattt> *mongodb
16:28:13 <cloudnull> they have a "raise exception" on all things meta query
16:28:38 <cloudnull> and meta query is only supported when using mongo
16:29:27 <cloudnull> i'd have to look at what tempest tests in terms of ceilometer, but if it calls something that does a meta query it would be a no go.
16:29:55 <palendae> Does horizon have integration tests to ceilometer taht we should exercise?
16:30:24 <mattt> doesn't look it palendae
16:30:24 <cloudnull> palendae no, but if it was deployed for testing using the AIO and wanted to see it working or tool around with it we'd run into problems.
16:30:56 <cloudnull> its also a completely different code path/driver within ceilometer.
16:31:33 <d34dh0r53> ^ makes me think that testing w/ mysql would not be a valid test
16:31:59 <palendae> Right
16:32:15 <andymccr> but thats a difference between testing that ceilometer deployed and testing ceilometer itself right?
16:32:22 <andymccr> like surely its ceilometers job to test the codepath with mongo
16:32:22 <cloudnull> we could get it in using mysql, it would just not be wants really used with its deployed; should someone choose to deploy it in prod. which is kinda against what / how we've tested all of the other openstacky things .
16:33:09 <andymccr> i agree with that, im just thinking if mongo getting in is a harder task as a first step it isnt a bad one, but its a non-event if the tests wont ever pass based on the meta query issue.
16:33:10 <cloudnull> but for the purpose of the spec it might be ok.
16:33:19 <palendae> It's close to how we've tested networking - neutron doesn't cover tests with the linuxbridge code path right now
16:33:21 <palendae> At least not in scenarios
16:33:50 <cloudnull> palendae thats true but we're exercising it using other scenarios .
16:33:53 <alextrcitiy> I think i'll just make a gate script that deploys a simple mongodb server with ceilometer. Then use that mongo server to test ceilometer
16:34:12 <palendae> Yes
16:34:19 <stevelle> that simple gate script could use the galaxy role previously identified.
16:34:30 <cloudnull> ^ that im good with
16:34:44 <alextrcitiy> Cool. I'll add that to the bp
16:36:06 <cloudnull> thanks alextrcitiy
16:36:29 <cloudnull> Sam-I-Am palendae https://review.openstack.org/#/c/173155/
16:36:48 <palendae> Yeah, I need to update that with suggestions people have made
16:37:10 <palendae> I do have some CRs related to it, but I'd like to do those slowly - first get in the directory with the sphinx config/make file, then build on that
16:37:40 <cloudnull> +1
16:38:02 <palendae> I have a first draft of a playbooks overview, too
16:39:15 <Sam-I-Am> palendae: wooo docs
16:39:16 <palendae> So, not really much to add to that right now
16:39:30 <palendae> Sam-I-Am: And docs that *we* should be writing
16:39:39 <palendae> Not "here's code write about it"
16:39:50 <Sam-I-Am> yeah. there will be some discussion about this on tuesday i think.
16:40:07 <Sam-I-Am> part of the overall "how do we get all the docs upstream"
16:40:25 <palendae> My goal with this is not all the docs, btw
16:40:25 <cloudnull> nice
16:40:37 <Sam-I-Am> palendae: er?
16:40:38 <palendae> But mostly for someone contributing to osad to get up to speed
16:40:49 <palendae> This specific BP
16:40:56 <Sam-I-Am> yes
16:41:05 <Sam-I-Am> but docs people might be touching this stuff
16:41:06 <palendae> There's still room for deployer docs
16:41:10 <palendae> Yes, true
16:41:17 <Sam-I-Am> diagrams, after all :)
16:41:50 <cloudnull> https://github.com/stackforge/os-ansible-deployment/blob/master/development-stack.rst#diagram-of-stack < best diagram ever!
16:42:08 <Sam-I-Am> cloudnull: until i make it better :)
16:42:26 <Sam-I-Am> needs more emoji
16:42:30 <Apsu> That's a fantastic diagram.
16:42:37 <cloudnull> So i think that we can close out on this one https://review.openstack.org/#/c/168976/ as it is
16:42:52 <Apsu> Needs some libaa or libcaca to render it
16:43:17 <Sam-I-Am> cloudnull: close out?
16:43:18 <cloudnull> the policy copy mechanism is in place and we can build on that to make that spec more supported / capable.
16:43:26 <palendae> +1
16:43:31 <palendae> Superseded
16:43:36 <cloudnull> ^ that one.
16:43:41 <Sam-I-Am> ahh
16:43:58 <palendae> Policy copying done, what we need next is ini updating
16:44:07 <cloudnull> yup
16:44:19 <cloudnull> we should look at creating a new spec for that extension .
16:45:21 <cloudnull> ok, well ill do that .
16:45:39 <cloudnull> #action item cloudnull create a new spec to extend copy_update module.
16:46:19 <cloudnull> #topic Open discussion
16:46:40 <cloudnull> open up the festivus !
16:47:07 <palendae> Does anyone know if our gate job has to inlcude 'dsvm' in it?
16:47:22 <cloudnull> http://festivusweb.com/images/seinfeld-festivusjpg.jpg
16:47:31 <palendae> os-ansible-deployment-dsvm-check-commit
16:47:35 <cloudnull> palendae no, we could take the name out i think
16:47:36 <palendae> From what I gather
16:47:41 <palendae> dsvm is DevStack VM
16:47:49 <palendae> Which is inaccurate in our case
16:48:04 <sigmavirus24_awa> the idea is the same though (an AIO with core services in it)
16:48:09 <palendae> Right
16:48:10 <cloudnull> thats something we can update/change in project config
16:48:10 <Sam-I-Am> we should just use devstack
16:48:16 <palendae> It's just not devstack
16:48:19 <sigmavirus24_awa> ==Sam-I-Am
16:48:28 <Sam-I-Am> then everything would work
16:48:30 <palendae> cloudnull: Yeah, I wanted to ask before I put a PR in for that
16:48:31 <cloudnull> # /kick Sam-I-Am
16:48:55 <cloudnull> :)
16:48:57 * Sam-I-Am feels loved
16:49:00 <palendae> We are doing the same thing, but it's not devstack (nor a VM, I don't think), so it's misleading
16:49:00 <cloudnull> palendae that should be fine
16:49:10 <cloudnull> you want to go bang that out ?
16:49:16 <palendae> Yeah
16:49:19 <sigmavirus24> palendae: you go and bug infra about it
16:49:35 <palendae> sigmavirus24: I will
16:49:37 <cloudnull> #action palendae to fix our gating name.
16:50:30 <cloudnull> who wants the fetivus pole next ?
16:50:30 <cloudnull> Bjoern__ where you be ?
16:50:30 <cloudnull> :)
16:51:42 <cloudnull> so are we done here?
16:52:10 <Sam-I-Am> perhaps
16:52:14 <cloudnull> i think so :)
16:52:15 <b3rnard0> kill it
16:52:19 <cloudnull> #endmeeting