Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2014-06-23

| Channels | #puppet-openstack index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:04 openstackgerrit Ricardo Rocha proposed a change to stackforge/puppet-ceph: add ceph-extras and fast-cgi repositories.  https://review.openstack.org/96065
01:16 britthouser joined #puppet-openstack
01:20 britthouser joined #puppet-openstack
01:50 xingchao joined #puppet-openstack
02:06 openstackgerrit Ricardo Rocha proposed a change to stackforge/puppet-ceph: add ceph-extras and fast-cgi repositories.  https://review.openstack.org/96065
02:08 mjblack joined #puppet-openstack
02:08 comptona joined #puppet-openstack
02:55 openstackgerrit Xav Paice proposed a change to stackforge/puppet-neutron: Update declaration of integration bridge  https://review.openstack.org/101790
03:09 gildub joined #puppet-openstack
03:18 openstackgerrit Xav Paice proposed a change to stackforge/puppet-neutron: Update declaration of integration bridge  https://review.openstack.org/101790
03:22 xavpaice gah! apologies for the spam
03:27 mjblack joined #puppet-openstack
03:37 fandi joined #puppet-openstack
04:16 bkero joined #puppet-openstack
04:34 daneyon joined #puppet-openstack
05:01 daneyon_ joined #puppet-openstack
05:10 ricardorocha joined #puppet-openstack
06:07 dachary ricardorocha: re-run gerritexec, vagrant was hang
06:07 dachary ricardorocha: I did the re-run just now ;-)
06:47 paraa joined #puppet-openstack
06:48 paraa This merge introduced a circular dependency for me: https://github.com/stackforge/puppet-neu​tron/commit/9518b390f0dac12612e7385a1e9  --> "(Neutron_plugin_ml2[ml2/mechanism_drivers] => Service[neutron-server] => Class[Neutron::Server] => Class[Neutron::Plugins::Ml2] => Neutron_plugin_ml2[ml2/mechanism_drivers])" - very strange
06:53 ddmitriev joined #puppet-openstack
06:57 artem_panchenko joined #puppet-openstack
06:58 gcha joined #puppet-openstack
06:58 EmilienM good morning
07:03 aldavud joined #puppet-openstack
07:09 arnaud_orange joined #puppet-openstack
07:23 arnaud_orange morgning
07:27 tfz joined #puppet-openstack
07:38 qba73 joined #puppet-openstack
07:49 mmagr joined #puppet-openstack
08:27 qba73 joined #puppet-openstack
08:31 aldavud_away joined #puppet-openstack
08:41 sileht joined #puppet-openstack
08:48 eugene_n joined #puppet-openstack
08:54 ddmitriev joined #puppet-openstack
09:04 eugene_n left #puppet-openstack
09:05 eugene_n joined #puppet-openstack
09:10 aldavud_ joined #puppet-openstack
09:44 mmagr joined #puppet-openstack
09:59 dachary bonjnour
10:43 comptona joined #puppet-openstack
10:59 britthouser joined #puppet-openstack
11:25 gildub joined #puppet-openstack
11:56 dmsimard aldavud_,dachary, xarses: Hopefully finishing my big production push this week, I'll have more time soon.
11:57 dmsimard Sorry about that. Crazy deadlines :)
11:57 dachary cool :-) good luck with that
11:59 dmsimard Actually doing a puppet overdose atm.
11:59 dmsimard :P
11:59 dmsimard Only wish it was on Ceph !
12:05 dachary is there anyone here involved in packstack ? I'd like to better understand where it is going with regard to the standard puppet openstack modules.
12:15 aldavud_ dmsimard: good luck :)
12:16 imcsk8 joined #puppet-openstack
12:19 imcsk8 joined #puppet-openstack
12:38 britthouser joined #puppet-openstack
12:39 britthouser joined #puppet-openstack
12:42 EmilienM mmagr: hey, dachary wanted me to introduce Red Hat people working on Puppet
12:42 EmilienM mmagr: so I know you and francesco, who is offline atm
12:43 dachary mmagr: hi :-)
12:44 dachary EmilienM: thanks for the introduction.
12:45 EmilienM my pleasure
12:45 EmilienM dachary: they are the most active for know
12:46 dachary mmagr: I'm curious to learn more about the redhat approach to puppet / openstack / ceph etc. Because I'm a recent hire and eager to better know what my colleagues are up to ;-)
12:46 gildub joined #puppet-openstack
12:59 dprince joined #puppet-openstack
13:07 tfz joined #puppet-openstack
13:11 tfhartmann joined #puppet-openstack
13:13 mmagr EmilienM, Yup, on this channel it's just me and Francesco
13:13 EmilienM cool
13:14 mmagr dachary, currently we don't work on Ceph at all unfortunately
13:16 blentz joined #puppet-openstack
13:23 dachary mmagr: maybe that's something I could help with ? https://github.com/ceph/puppet-ceph has been a hobby so far but there is no reason why it could not be part of my work ;-)
13:34 morazi joined #puppet-openstack
13:58 Fabien joined #puppet-openstack
14:05 dachary xarses: dmsimard aldavud_ : the puppet-ceph integration machine is moving to another datacenter this week. There will be a one or two days downtime. Isn't it nice to know there will be a downtime in advance for once ? ;-)
14:06 aldavud_ dachary: thank you for the heads up :)
14:06 aldavud_ talking of integration tests
14:07 thartmann joined #puppet-openstack
14:07 aldavud_ I am currently rerunning the tests on all the profile tests and it looks good so far
14:08 aldavud_ but it is somewhat conflicting with your change to run in parallel
14:09 aldavud_ so we should probably make them dependent and adapt the tests as well
14:10 aldavud_ or merge one change first and then adapt the other accordingly
14:16 aldavud_ dachary: ^
14:16 mtl1 joined #puppet-openstack
14:17 dachary you can ignore my // changes for now : I'll resolve the conflicts. Thanks for the warning though ;-)
14:21 prad_ joined #puppet-openstack
14:22 mmagr joined #puppet-openstack
14:24 openstackgerrit Pradeep Kilambi proposed a change to stackforge/puppet-neutron: Support for Cisco ML2 Mech Driver  https://review.openstack.org/101338
14:30 mjblack joined #puppet-openstack
14:31 mjblack joined #puppet-openstack
14:41 hogepodge_ joined #puppet-openstack
14:42 cwolferh joined #puppet-openstack
14:50 EmilienM meeting?
14:50 EmilienM hogepodge: ^
14:50 EmilienM oops
14:50 EmilienM i'm too early
14:52 michchap EmilienM: do you have a patch for puppet-openstack-cloud adding ml2 support using the new patch?
14:52 EmilienM michchap: not yet
14:52 EmilienM i can do it this week
14:53 michchap EmilienM: No rush :)
14:54 xingchao joined #puppet-openstack
14:55 rwsu joined #puppet-openstack
14:56 aldavud joined #puppet-openstack
14:57 mmagr joined #puppet-openstack
14:57 britthouser joined #puppet-openstack
14:59 hogepodge Yup, give me a few minutes.
14:59 hogepodge One minute :-)
14:59 hogepodge Meeting notes. https://etherpad.openstack.org​/p/puppet-openstack-2014-06-23
15:00 hogepodge Hello everyone!
15:01 hogepodge #start meeting
15:01 EmilienM michchap: should we talk about world cup and Australia team? :-P
15:01 hogepodge You should add that to the agenda
15:01 michchap EmilienM: I will be supporting France
15:01 EmilienM \o/
15:02 hogepodge #topic puppet-vswitch
15:02 sbadia hi
15:03 hogepodge This is kind of a short one.
15:03 hogepodge RedHat has been sending patches to vswitch, but they’ve been languishing a bit
15:03 michchap this is just about gildub's patch?
15:03 arnaud_orange hi
15:04 hogepodge Yeah.
15:04 aldavud hi
15:04 hogepodge Just wanted to make sure core was following the project and had the resources to do reviews (myself included)
15:04 hogepodge The infra team had concerns that the project would fork (since there were already so many forks out there)
15:05 michchap the only thing that worries me is that the keep_ip behavior becomes default for redhat and not for debian
15:05 hogepodge That was one of my concerns with the patches too, that Debian was not being tested or keeping feature parity.
15:06 michchap I don't want to push the responsibility for keeping the two in sync solely onto redhat because it's not really their problem, but I can see it being a bit confusing if they're diverging.
15:06 hogepodge is gildub or ivan here?
15:06 EmilienM nobody from RH here afi
15:06 EmilienM afik
15:07 hogepodge Not just confusing, but possibly breaking.
15:07 mdorman joined #puppet-openstack
15:07 hogepodge Are there any ubuntu experts here who would want to volunteer to help keep parity?
15:08 dachary EmilienM: we're from redhat I'm afraid ;-)
15:09 EmilienM dachary: I mean, someone who worked on it
15:10 hogepodge This week I’m going back to working on Ubuntu, and can try to bring it back in line. I’m definitely weaker there than on RH, though.
15:12 michchap all my testing is on centos so I'm not much use atm.
15:13 hogepodge Ok, I have to work on both anyway, so I’ll suck it up. ;-)
15:14 hogepodge #topic integration tests
15:14 hogepodge This was proposed last week, but we didn’t get to it.
15:15 hogepodge Who had put this forward?
15:15 aldavud dachary did afaik
15:16 hogepodge Yeah, just saw that. dachary?
15:16 dachary :-)
15:16 dachary shortly : here is what we do for puppet-ceph :
15:16 dachary https://github.com/stackforge/puppet-ceph/bl​ob/master/spec/system/ceph_mon_spec.rb#L319
15:17 dachary is used when working on part of puppet-ceph and it runs a deployment based on rspec-system (deprecated) locally
15:17 dachary then when we push, the puppetceph bot picks up and runs all of the tests
15:17 dachary and reports back https://review.openstack.org/#/c/100522/
15:17 dachary (see at the end)
15:18 dachary it is deployed on a bare metal and triggered by https://pypi.python.org/pypi/gerritexec
15:18 dachary we're moving to beaker and docker
15:18 dachary I'm *very* interested to hear about any other mean of integration tests people are using :-)
15:19 hogepodge Mine are still crude. Vagrant on my box, playing with beaker, but right now we don’t have the isolated infra to trigger tests in to
15:19 crinkle joined #puppet-openstack
15:19 dvorak I think beaker makes sense, is the proposal that someone run hardware to test the other modules also?
15:20 openstackgerrit Pradeep Kilambi proposed a change to stackforge/puppet-neutron: Support for Cisco ML2 Mech Driver  https://review.openstack.org/101338
15:20 michchap mine is all vagrant, though I was doing it with heat last year.
15:20 hogepodge Mirantis said they have infra in place.
15:21 dachary beaker is appealing because it can be used with docker and therefore not require a bare metal (we have an OpenStack instance setup for that but it's not used yet so I could not tell how well it performs).
15:21 hogepodge We’re going to start adding OpenStack support to Beaker too.
15:21 dachary hogepodge: looking forward to that :-)
15:22 dachary but it's not a blocker
15:22 hogepodge I’m migrating my tiny hardware installaiton here to give them a test platform.
15:22 hogepodge (havana -> icehouse)
15:22 dachary michchap: you have a single hardware to run vagrant based tests ?
15:22 hogepodge dprince had smokestack running at RH, but he’s been on Triple-O and that system has been down since January.
15:23 michchap dachary: I test everything on my desktop right now. I bought a lot of RAM so I can do a 3 node HA cluster with 2 compute nodes. We're fixing our internal cloud up atm so I'll move back to heat very soon.
15:23 dachary if people want to give beaker + docker integration tests a try, there are instructions and a working example at https://github.com/elasticsearch/puppet-e​lasticsearch/blob/master/CONTRIBUTING.md
15:24 dachary hogepodge: that's bad news :-(
15:24 dachary EmilienM: I think you have an integration platform based on jenkins, right ?
15:24 EmilienM dachary: our CI is not exposed on public network for now
15:24 hogepodge Yeah, I asked him about it before the summit. He just hasn’t had time.
15:25 dachary EmilienM: I don't mean the platform itself but the configuration
15:25 hogepodge Is there any way of taking advantage of the OS infra resources?
15:25 EmilienM yes
15:25 EmilienM but we don't use vagrant
15:25 EmilienM and we run multi nodes, that would be overkill to start
15:25 dachary EmilienM: whatever works, it would be nice to have the configuration published as an example of how to do integration tests
15:26 hogepodge dachary with Tempest config too, if anyone is using it. I’m always curious to see which tests people are skipping.
15:27 EmilienM dachary: the wizard would be a vagrant script with basic manifests
15:27 hogepodge My experience is a lot of Tempest is broken for a variety of reasons.
15:27 EmilienM and tempest running after that
15:27 dachary EmilienM: how much of it passes ?
15:27 aldavud EmilienM: do you cover multiple distros?
15:27 EmilienM aldavud: yes, Debian/Ubuntu/RH
15:28 EmilienM dachary: they all pass afik
15:28 dachary EmilienM: are you provisioning machines with heat ?
15:29 EmilienM dachary: yes it's public, let me find it
15:29 EmilienM the heat template is here: https://github.com/enovance/openstack-yaml-in​fra-3nodes-fullha/blob/master/heat.yaml.tmpl
15:30 EmilienM it's an example of course
15:30 daneyon joined #puppet-openstack
15:31 dachary hogepodge: that's all I have on the topic of integration tests, thanks for adding it to the agenda ;-)
15:31 dvorak seems like the big advantage of beaker would be that people can run the tests on their laptop/dev box/etc fairly easily.
15:31 xingchao we are using Jenkins & internal openstack platform to run integration test
15:32 hogepodge xingchao also using heat?
15:32 dachary xingchao: cool, are the configuration files publicly available ?
15:32 xingchao hogepodge:  no, using shell script :)
15:33 hogepodge :-)
15:33 xingchao dachary: no yet, I will plan to publish it on github
15:33 dachary thanks :-)
15:35 hogepodge #topic 4.1.0 release
15:35 hogepodge Another short one,
15:35 hogepodge I cut a release last Friday, mostly bug fixes but some additional APIs
15:36 hogepodge When I started doing this I was asked to just merge and not bother with the 2x+2 reviews. Just wanted to make sure we’re still ok with that.
15:36 hogepodge I like to leave the reviews up for a bit so people can comment though, like sbadia helping out with a feture I missed
15:37 sbadia argee for me, sorry i wasn't aware of this, but indeed (for spam)
15:37 hogepodge sbadia no, it was helpful, I’m glad you caught it.
15:38 hogepodge Ok, moving on?
15:38 sbadia yup
15:38 hogepodge #topic puppet-openstacklib first steps
15:39 hogepodge puppet-openstacklib and puppet-openstack_extras were merged into Stackforge last week.
15:39 hogepodge Should add them to you watched projects
15:39 EmilienM let's start the work now :)
15:39 hogepodge Setting up the projects is the first step now, along with initial code
15:40 hogepodge I was thinking a good first step would be adding repository code to lib
15:40 EmilienM I don't see puppet-openstack_extras
15:40 hogepodge Ugh, you’re right. extras isn’t there yet.
15:40 hogepodge My mistake.
15:40 hogepodge lib is though.
15:41 hogepodge I’ll poke infra
15:41 hogepodge Any other easy tasks to start populating lib with?
15:42 hogepodge We already have one spec for configs https://review.openstack.org/#/c/99​509/1/specs/openstack_configs_class.rst,cm
15:42 michchap common endpoint type would be easy
15:42 hogepodge +1 on that
15:42 hogepodge (common endpoint)
15:43 michchap to replace this and the others like it: https://github.com/stackforge/puppet-neutr​on/blob/master/manifests/keystone/auth.pp
15:44 hogepodge Any thoughts on lib organization? We are free to make choices and change things around early on, but once we hit Juno release it’s going to be harder to shift the API around
15:45 hogepodge might be good to spec something out
15:47 hogepodge So repos, endpoints, any others?
15:47 michchap I think just let things flow in there as people need them.
15:47 xingchao db
15:47 michchap +1 db
15:47 xingchao mq
15:48 michchap +1 mq
15:49 hogepodge great.
15:49 hogepodge #topic new devs and operators
15:50 hogepodge We talked about giving space to new devs and operators to ask questions during the meeting. I’ll open that up now for the last 10 minutes.
15:51 hogepodge I’ll also take the opportunity to introduce Colleen, crinkle here, who’s a new intern at Puppet Labs.
15:51 EmilienM welcome crinkle !!
15:51 hogepodge She’ll be doing some library work with aviator starting in a few weeks.
15:52 crinkle o/
15:52 dvorak awesome
15:52 JayJ joined #puppet-openstack
15:53 xingchao ^^
15:54 eugene_n Hi. I am new one here and I have small question. I tested puppetlabs/openstack to deploy openstack icehouse on ubuntu 14.04  and got some issues. i.e. Ubuntu is using ml2 plugin by default. Are there any plans to switch from openvswitch-plugin to ml2?
15:55 aimon very cool! (aviator work)
15:55 hogepodge eugene_n both openvswitch-plugin and ml2 should be using the same ovs system.
15:55 hogepodge I’m planning on switching it over, and the 4.1.0 modules added ml2 support.
15:56 doo joined #puppet-openstack
15:56 hogepodge I’ve been struggling with it a bit, though. I haven’t been alone in that trouble with ml2, still debugging on my side.
15:56 hogepodge I think others have had more success, though.
15:57 EmilienM :q
15:58 eugene_n I was able to make it work , I can provide patch. But it was tested only in ubuntu and as I undestood you are mainly developing on centos
15:59 hogepodge eugene_n I’m biased, but don’t want to be.
15:59 hogepodge yes, patches welcome.
15:59 hogepodge Please send.
16:00 eugene_n Ok.Thanks
16:01 hogepodge thank you.
16:02 hogepodge 9:00, so I’ll close the meeting, but still around for questions. Have a fantastic week everyone!
16:02 hogepodge #end meeting
16:02 eugene_n left #puppet-openstack
16:03 EmilienM michchap: good luck :P
16:03 michchap EmilienM: thanks.
16:09 starmer joined #puppet-openstack
16:15 dvorak could I get some eyes on this review? It fixes neutron resources so that it works if multiple regions exist - https://review.openstack.org/#/c/99801/
16:35 hogepodge dvorak testing
16:36 JayJ joined #puppet-openstack
16:49 rmoe joined #puppet-openstack
16:52 aimon_ joined #puppet-openstack
16:52 dvorak hogepodge: thanks :)
17:00 mfisch mdorman: did you ever push your "specify a cert" change for puppet-keystone (PKI)?
17:02 mdorman i had to resubmit it as a new change… one sec, let me find it.
17:02 mdorman https://review.openstack.org/#/c/101260/
17:03 mdorman i added the reviewers from the previous change to this one, but there’s been no reviews yet
17:04 aldavud joined #puppet-openstack
17:04 mfisch mdorman: thanks, we're punting on PKI for awhile but having this in place will help us when we use it
17:04 mdorman cool
17:05 mdorman we will probably look at pki tokens again once there is token compression, too
17:05 mfisch there is PKIZ but not in Icehouse yet
17:06 mfisch mdorman: also this idea to not store anythning in the DB re: tokens: https://review.openstack.org/#/c/9597​6/13/specs/juno/non-persistent-tokens.rst
17:06 mfisch mdorman: PKI tokens dont solve any DB Scaling issues if they have to be stored
17:07 mdorman huh, good point.  yeah i’ll take a look at that spec
17:07 mfisch sounds like in Juno tokens will be rainbows and unicorns
17:07 mdorman :)
17:07 mdorman from what i heard, *everything* is rainbows and unicorns in Juno :)
17:08 mfisch they got tired of re-writing it so it's actually ${next_release}
17:12 xarses joined #puppet-openstack
17:12 hogepodge bodepd I was hoping you’d rebase that. Project moving was us trying to accomplish the same thing I think
17:16 mfisch mdorman: reviewed, mostly looks goosd
17:17 mdorman thanks
17:18 hogepodge dvorak lgtm
17:19 xarses hogepodge: we will add CI to puppet-openstack as we start tracking the upstream for each of the modules. our progress for the latter is trackable on https://blueprints.launchpad.net/fuel​/+spec/merge-openstack-puppet-modules I don't have anything formal to share with the prior yet.
17:19 mfisch man I completely spaced on the meeting today
17:20 hogepodge I sent the reminder late mfisch
17:21 mfisch monday mornings are bad for me as I avoid looking at my work cal all weekend
17:23 mmagr joined #puppet-openstack
17:23 xarses hogepodge, dachary : for infra we use jenkins and libvirt (via https://github.com/stackforge/fuel-devops) and our ci jobs in https://github.com/stackforge/fu​el-main/tree/master/fuelweb_test at the moment. Currently for puppet (fuel-library) we just use one job for each distro to deploy a complete env
17:32 dachary xarses: thanks for sharing :-)
17:32 xarses sorry i missed the meeting
17:34 hogepodge xarses thanks@
17:34 hogepodge s/@/!
17:35 tfz joined #puppet-openstack
17:36 xarses dachary: hogepodge here is an example review of our fuel-library CI https://review.openstack.org/#/c/86008/
17:36 dachary cool
17:37 xarses IIRC centos is one controller, one compute with nova networks, and ubuntu is ha 3 controller and 1 compute neutron vlan
17:50 markvoelker1 joined #puppet-openstack
17:51 markvoelker1 left #puppet-openstack
18:20 openstackgerrit Bob Callaway proposed a change to stackforge/puppet-cinder: Add secret flag to netapp_sa_password option  https://review.openstack.org/101971
18:21 mfisch hey hogepodge or Hunner for non-openstack puppet changes is a github pull request sufficient?
18:21 hogepodge If it’s not in Stackforge, yes
18:21 mfisch thx
18:25 openstackgerrit Bob Callaway proposed a change to stackforge/puppet-cinder: Add secret flag to netapp_sa_password option  https://review.openstack.org/101971
18:29 csschwe joined #puppet-openstack
18:36 mfisch hogepodge: this is openstack related but not in stackforge: https://github.com/puppetla​bs/puppetlabs-ntp/pull/170
18:36 hogepodge That’s the right place to do it.
19:03 nibalizer hogepodge: so you probably answer this 10x a day
19:03 nibalizer but which is the official puppet modules?
19:03 mfisch I thought you were going to ask where he buys his sport coats
19:03 nibalizer the puppet-* on stackforge or puppetlabs-openstack ?
19:03 nibalizer also the puppetlabs-openstack pulls in puppetlabs-nova for example, where is that?
19:04 mfisch I assume it's stackforge
19:04 mfisch thats what we're using anyway
19:05 hogepodge nibalizer I have a draft e-mail about this.
19:05 nibalizer okay cool
19:05 nibalizer who is it to?
19:05 hogepodge puppet-openstack on Stackforge is suffering pretty badly from code rot. There’s been some effort to maintain it, but not really that much as everyone uses their own composition layers
19:05 nibalizer is there a ML i should probabaly be on?
19:06 hogepodge the puppetlabs-openstack module is my own composition layer, I use it for development and testing. It’s lacking things like HA, which is problematic as a production deployment module. It’s the one that Puppet Labs is supporting
19:07 hogepodge We talked about replacing puppet-openstack with it, but right now functionality is being moved over to puppet-openstacklib
19:07 hogepodge nibalizer puppet-openstack on Google Groups https://groups.google.com/a/puppetlab​s.com/forum/#!forum/puppet-openstack
19:08 nibalizer ok
19:08 nibalizer is there secret sacuse that puppetlabs hasn't open sourced for HA and stuff?
19:08 hogepodge I would very much for all of the modules hosted on Stackforge to be released in a Puppet Forge Stackforge namespace, largely because that reflects the development of the modules themselves.
19:08 dvorak mfisch: what we're using is a horrible mismash of stackforge openstack, coi and coe.
19:09 hogepodge no secret sauce, I haven’t written it yet.
19:09 nibalizer hogepodge: so the puppet-* modules on stackforge are released as e.g. puppetlabs-nova on the forge?
19:09 hogepodge The sauce isn’t really secret from anyone else. See, for example puppet-openstack_builder and puppet-openstack_cloud
19:10 hogepodge nibalizer, correct. For Juno I want that to be stackforge-nova, etc.
19:10 nibalizer okay
19:10 hogepodge Because those modules are community developed and supported by a large number of companies, not just Puppet Labs.
19:10 nibalizer whats the plan for beaker tests?
19:10 hogepodge I think the current state doesn’t reflect reality.
19:11 hogepodge We’re building support for openstack into beaker.
19:11 hogepodge I’m updating a small cluster for it
19:11 hogepodge But we need to work out how to create a standalone external CI system for community projects so we can integrate Gerrit.
19:11 nibalizer oh
19:12 nibalizer so you would run a third-party ci system
19:12 hogepodge That involved other groups in the company, and is a larger initiatlive.
19:12 hogepodge I just test manually in Vagrant right now
19:12 nibalizer run the beaker tests over there, then report on the reviews
19:12 hogepodge Yes
19:12 nibalizer i work with the -infra team and am scratching my head on how to add beaker tests to the puppet code we use over there
19:12 hogepodge We can’t have automated untrusted code running internally.
19:13 hogepodge nibalizer I’d like to help
19:13 nibalizer so i have two ideas
19:13 nibalizer which are both kinda janky
19:13 nibalizer and focus on the problem that beaker wants to be the controller, it wants to provision nodes and reset them and all that
19:13 nibalizer and currently there is other machinery for doing that so the two are in conflict
19:14 hogepodge ok
19:14 nibalizer my one idea is that recently, because of trippleo, you can now tell nodepool you need multinodes
19:14 hogepodge +1 to that
19:14 nibalizer so we could have beaker run on one node, and it could configure and test against the other nodes
19:14 hogepodge How recent is it?
19:14 nibalizer le shrug
19:14 nibalizer the other is electrical added docker support
19:14 anode joined #puppet-openstack
19:14 nibalizer and there are trusty nodes in nodepool now
19:15 nibalizer so we could bring up a single trusty node, run beaker on that, then it could provision all the docker containers it wanted
19:15 dvorak will docker work for things like puppet-vswitch though?
19:15 nibalizer the problem with that is things like iptables and stuff don't work in an LXC context
19:15 dvorak nod
19:15 nibalizer dvorak: probably not
19:16 nibalizer but i do think that docker is probably a pretty good 80/20 split in terms of getting a bunch of tests automated
19:16 hogepodge docker doesn’t abstract away network resources?
19:16 nibalizer and we can design complicated systems for the other stuff
19:17 nibalizer hogepodge: i'm not really a docker person, i've used the lxc utilities extensively for years tho
19:17 nibalizer and those fall flat
19:17 nibalizer also lxc can't do anything that requires kernel namespacing so nfs and iscsi are out
19:19 hogepodge as well may be ovs
19:20 hogepodge anode, could beaker run ‘fake’ provisioning that can just find existing machines?
19:20 anode i believe that you can provide hostnames and ips and run with no-provision it will just connect
19:20 anode i haven’t tested that workflow in a while, but it’s there
19:21 hogepodge nibalizer anode is our beaker guru
19:22 nibalizer so i guess i wonder if that would be sufficent
19:22 nibalizer like if i have separate context do .... stuff ... end blocks, does beaker reprovision in that?
19:22 nibalizer since nodepool won't reprovision for you
19:23 nibalizer however, what some people are doing is freezing the lvm block device, running tests, then kexecing to refresh when its time to run more tests
19:23 nibalizer there is also a union filesystem in there somewhere to allow the tests to run normally
19:24 anode no, beaker assumes a single provisioning attempt at the beginning of a run, and a single shutdown/terminate at the end
19:24 anode though we are looking to integrate the ability to revert to snapshot before/after tests - but that would be up to the beaker test to call
19:24 hogepodge nibalizer do you have an example somewhere of how infra provisions multiple nodes? What that lifecycle looks like?
19:25 nibalizer no, i've only been told that you can do it
19:25 nibalizer i can go looking i think i know where to look
19:26 hogepodge ok
19:29 xarses nibalizer: hogepodge, snapsharts are part of what the fuel CI tests are doing (but with libvirt as the snap) to help save time
19:29 xarses snapsharts/snapshots
19:30 hogepodge nibalizer I’m hoping that there’s an easy way to use noprovision to handle this
19:30 nibalizer that would be ideal i think
19:48 nibalizer hogepodge: so update,  no one uses the multinode yet
19:48 nibalizer but if you requisition a node of this type: http://git.openstack.org/cgit/openstack-in​fra/config/tree/modules/openstack_project/​templates/nodepool/nodepool.yaml.erb#n104 you'll get nodes
19:49 nibalizer and if you read that multinode_setup.sh script it has all the ip addresses of the sub nodes in a file in /etc/nodepool/sub_nodes, and it already has ssh key trust between the primary node and the sub nodes
19:49 nibalizer in other words, i think this is going to work
19:49 mjblack_ joined #puppet-openstack
19:50 mjblack joined #puppet-openstack
19:50 xarses dmsimard: nice write up on Tengine
19:51 hogepodge nibalizer that would be fantastic. I’m going to drop those links into the meeting notes re ci
19:51 nibalizer okay
19:51 nibalizer im gonna try first on openstack-nova just for grins, unless you have a different module you'd reccomend I try using as a test case
20:07 hogepodge :-)
20:07 anode hogepodge: looking over the beaker code you can use a hypervisor of ‘none’ and it will just use the provided hostname/ip for connecting
20:08 hogepodge nibalizer ^ excellent. Thanks!
20:13 dmsimard xarses: Hey, thanks
20:13 dmsimard xarses: Mirantis even mentioned it on twitter, achievement unlocked :D
20:15 xarses =)
20:19 aimon_ joined #puppet-openstack
20:22 mfisch The puppet-horizon defaults let passwords be logged
20:22 mfisch FYI
20:22 mfisch I'll see about fixing
20:26 dachary bodepd: are you around by any chance ?
20:29 openstackgerrit Pradeep Kilambi proposed a change to stackforge/puppet-neutron: Support for Cisco ML2 Mech Driver  https://review.openstack.org/101338
21:23 mdorman_ joined #puppet-openstack
21:26 mdorman_ joined #puppet-openstack
21:36 openstackgerrit Spencer Krum proposed a change to stackforge/puppet-nova: Initial rspec-beaker testing scaffolding  https://review.openstack.org/102020
21:44 openstackgerrit Carl Henrik Lunde proposed a change to stackforge/puppet-nova: Do not set special_service_provider on Red Hat  https://review.openstack.org/101775
21:51 openstackgerrit Matt Fischer proposed a change to stackforge/puppet-horizon: Switch default log handler level to INFO  https://review.openstack.org/102025
21:51 bodepd dachary: I was just about to start working
21:59 openstackgerrit Matt Fischer proposed a change to stackforge/puppet-horizon: Switch default log handler level to INFO  https://review.openstack.org/102025
22:15 britthou_ joined #puppet-openstack
22:31 hogepodge joined #puppet-openstack
22:54 aldavud joined #puppet-openstack
23:16 nibalizer so if im reading this correctly puppet-nova tries to use mysql::python which hasn't existed in the mysql module for some time
23:17 dvorak is this the thing where you have to set the version of the mysql module you're using?
23:18 dvorak I think it defaults to like 0.9
23:18 dvorak and you need to set it to 2.x
23:18 nibalizer well its installing 2.2 or 2.3 and mysql::python was last seen in 1.0.0 if im reading this right
23:19 dvorak nibalizer: https://github.com/stackforge/puppet-nova/​blob/master/manifests/db/mysql.pp#L38-L41
23:19 dvorak I think it'd be reasonable to change the default for juno
23:19 dvorak I don't think that nova is the only module with this problem
23:20 nibalizer dvorak: https://github.com/puppetlabs/pupp​etlabs-mysql/tree/2.2.x/manifests
23:20 nibalizer there is no mysql::python in 2.2
23:20 dvorak right, that's why it's not used if you tell it you're using 2.2
23:20 dvorak https://github.com/stackforge/puppet-nova/​blob/master/manifests/db/mysql.pp#L59-L78
23:23 nibalizer omg
23:23 nibalizer wat
23:23 * nibalizer jawdrop
23:25 britthouser joined #puppet-openstack
23:25 nibalizer im not even sure to use this
23:29 dvorak 0.9 came out 11 months ago, it's not really crazy that was the default for icehouse
23:29 nibalizer im suprised taht we pass in a module version
23:29 nibalizer not that im not getting over it
23:29 nibalizer just mind=blown
23:29 nibalizer i think i've figured out what to do
23:29 dvorak how else would you do it?
23:29 nibalizer you can pass in a mysql_module to the top nova class
23:29 dvorak yes
23:29 nibalizer dvorak: oh no i get it
23:29 nibalizer i dont have a better solution
23:30 dvorak I'll submit a review to change the default
23:30 nibalizer it doesn't blcok me
23:30 nibalizer so im not worried
23:30 dvorak I ran into the same thing a few months ago
23:30 nibalizer im wondering if we should create a private.pp in puppet modules to reproducte modulefile/metadata.json data but in a way that puppet could reach in and grab it
23:31 nibalizer that way you could check out $::mysql::private::version  to get the version of the module
23:32 dvorak you can't put that into 0.9 though
23:32 nibalizer nope
23:32 nibalizer im talking about the future
23:33 nibalizer and you could add it to 0.9.1
23:33 nibalizer since it wouldn't break anything
23:35 hogepodge markvoelker ping
23:36 nibalizer oh sweet i've got it sorta working
23:36 nibalizer dvorak: thanks for your help
23:36 dvorak np
23:36 hogepodge nibalizer default was 0.9, we can push it up to 2.2 for juno
23:37 dvorak hogepodge: apparently changing it breaks like 20% of the tests, somehow
23:37 hogepodge Yes, the tests need to be updated
23:37 nibalizer also there is a python.pp in tests/ on the mysql module taht can be nuked
23:37 hogepodge There was no way to reasonably test two versions
23:37 dvorak I'll try to get a review submitted for nova and we can do the other modules once that looks good
23:38 hogepodge Is anyone here use ml2 with ovs/gre?
23:38 hogepodge I feel like I’m banging my head against a wall here.
23:38 dvorak we're using ovs & vxlan
23:38 dvorak speaking of which, did emilien's patch ever get merged?
23:38 hogepodge What’s the difference between vxlan and gre? Is there an advantage to the former?
23:39 hogepodge It did, but it wasn’t backported
23:39 dvorak I definitely think it should be merged back to icehouse
23:39 dvorak lots of people (us) will want to go to ml2 before juno, to make upgrades easier
23:40 dvorak vxlan is actually designed to be a overlay networking protocol, gre isn't
23:40 dvorak gre works, but it's a bit more hackish
23:40 hogepodge It needs to go back, yes
23:40 hogepodge So ovs with vxlan?
23:41 dvorak I don't think many people will be doing new ml2/gre installs
23:41 hogepodge Do you have a sample puppet config I can take a look at?
23:41 hogepodge I don’t care what the network layer I use is, as long as it’s on ml2
23:41 dvorak ours is kind of spread out all over the place, I can definitely answer questions though, if that helps
23:42 openstackgerrit Clayton O'Neill proposed a change to stackforge/puppet-neutron: Configure OVS Agent when using ML2 plugin  https://review.openstack.org/102054
23:42 openstackgerrit Spencer Krum proposed a change to stackforge/puppet-nova: Initial rspec-beaker testing scaffolding  https://review.openstack.org/102020
23:43 nibalizer hogepodge: ping
23:43 dvorak let's see if that passes tests
23:43 nibalizer so i've created some simple beaker tests
23:43 hogepodge nibalizer pong
23:43 dvorak we're using a variant of that on icehouse, so I think it might
23:43 nibalizer so step 1 is done
23:43 nibalizer step 2/3 is a combined step where we add a nongating test to the puppet-nova projcet throught the infra project
23:43 nibalizer then hold on to our butts
23:44 hogepodge ok
23:44 nibalizer also anode how precisely do i specify a host/ip in the nodeset when hypervisor is None?
23:45 hogepodge I get ml2 networks build on CentOS 6, but my config bombs out when trying to start a VM and attach it to a private network
23:45 nibalizer hogepodge: so i think if I show up in #openstack-infra with random tests to add to the puppet-openstack repo they'll probably give me funny looks, so hopefully you can show up at that time and express that this is somethign the project wants
23:45 dvorak oh, I have to update the fixtures, that's why I'm getting so many failures
23:46 hogepodge Let me know when you head over.
23:47 anode nibalizer: the ip can just be ip: 00.00.00.00
23:47 anode nibalizer: could you paste me your hosts file?  i can check for correctness
23:48 nibalizer well right now i just have a vagrant based one
23:48 nibalizer trying to figure out what i need to do to change it
23:48 anode remove box & box_url
23:48 nibalizer http://paste.ubuntu.com/7692745/
23:49 anode change hypervisor: vagrant to hypervisor: none
23:49 nibalizer kk
23:49 anode and add an entry for ip: value
23:50 nibalizer ok
23:50 anode unless the box in question can be accessed by its name, then you don’t need ip
23:50 nibalizer i think we'll have ips
23:50 nibalizer but beaker will just ssh in as root with whatever ssh key that it has avialable?
23:50 dvorak heh, ok, with the fixture fixed, only two tests fail
23:51 anode yes, you’ll have to ensure that you can log in as passwordless root
23:51 nibalizer another question: is it possible to have multiple files hanging out in nodepool/* and choose which ones to use with an environment variable?
23:51 anode so you should create a new passwordless key and add it to the hosts
23:51 nibalizer i'd like to have the vagrant-runnable files next to the nodepool ones?
23:51 anode yeah, that is env settable
23:51 nibalizer ya the openstack multinode system sets all that up for me
23:51 anode are you using beaker or beaker-rspec?
23:51 nibalizer beaker-rspec or maybe rspec-beaker
23:51 anode okay
23:52 nibalizer yea beaker-rspec 2.2.4
23:52 anode here’s the env vars you’ll want https://github.com/puppetlabs/beake​r/wiki/How-to-Write-a-Beaker-Test-f​or-a-Module#supported-env-variables
23:52 anode i just released beaker-rspec 2.2.6 if you want to update
23:52 anode then it will have latest beaker
23:53 nibalizer okay aewsome
23:57 nibalizer ty
23:57 anode i’ll be heading out soon (hitting 5 o’clock around here)
23:58 anode but i can help more tomorrow
23:59 gildub joined #puppet-openstack
23:59 openstackgerrit Clayton O'Neill proposed a change to stackforge/puppet-nova: Change default MySQL module api to 2.x  https://review.openstack.org/102060

| Channels | #puppet-openstack index | Today | | Search | Google Search | Plain-Text | summary