17:00:22 <jroll> #startmeeting ironic 17:00:23 <openstack> Meeting started Mon Jan 11 17:00:22 2016 UTC and is due to finish in 60 minutes. The chair is jroll. Information about MeetBot at http://wiki.debian.org/MeetBot. 17:00:24 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:00:26 <openstack> The meeting name has been set to 'ironic' 17:00:30 <rpioso> o/ 17:00:30 <jroll> hi everyone 17:00:30 <TheJulia> o/ 17:00:32 <jlvillal> \o/ 17:00:34 <lucasagomes> hi there 17:00:35 <dtantsur> o/ 17:00:35 <rloo> hi 17:00:37 <NobodyCam> o/ 17:00:38 <mgould> o/ 17:00:44 <jroll> as always, agenda is here: 17:00:45 <jroll> #link https://wiki.openstack.org/wiki/Meetings/Ironic#Agenda_for_next_meeting 17:00:48 <krtaylor> o/ 17:00:52 <jroll> nice and light today :) 17:01:05 <devananda> o/ 17:01:13 <jroll> #topic announcements and reminders 17:01:26 <sambetts> o/ 17:01:30 <jroll> #info IPA releases this week, 1.1.0 from master and 1.0.1 from stable/liberty 17:01:34 <[1]cdearborn> o/ 17:01:41 <vdrok> o/ 17:01:44 <cinerama> o/ 17:02:09 <jroll> also, networks patches are very close, I'd love to get eyes on those 17:02:16 <jroll> anyone else have announcements? 17:02:24 <jlvillal> Gate? 17:02:34 <rloo> jroll: midcycle virtual whatever? 17:02:43 <jroll> ditto for manual cleaning, those have 2x +2 17:02:50 <jroll> rloo: let's talk in open discussion 17:02:58 <rloo> jroll: ok 17:02:59 <jroll> jlvillal: gate should be clear now afaik 17:03:03 <e0ne> jroll: I've got one from cinder side but I could hold it until open discussion 17:03:11 <jlvillal> Great! 17:03:18 <rloo> jroll: wasn't there a gate issue in stable/liberty or kilo? 17:03:38 <jroll> e0ne: yes pls 17:03:40 <dtantsur> gate is NOT clear yet 17:03:41 <lucasagomes> we still need to get the patches to allow dib to be tested from source in 17:03:45 <jroll> oh 17:03:51 <jroll> dtantsur: can you give more info please? 17:03:54 <dtantsur> but will be as soon as DIB pinning finally lands 17:03:57 <jroll> rloo: timeoutssssss 17:04:05 <dtantsur> the patch to g-r is in flight: https://review.openstack.org/#/c/265775/ 17:04:08 <jroll> dtantsur: oh, got it, thanks 17:04:28 <krotscheck> o/ 17:04:36 <dtantsur> after that please make sure we do land https://review.openstack.org/#/c/264736/ or we're broken again in the next release 17:04:40 <sergek> o/ 17:04:57 <dtantsur> Kilo is essentially dead due to constant timeouts 17:04:58 <sambetts> jroll: tinyipa has tested successfully in the gate, https://review.openstack.org/#/c/259089/ and https://review.openstack.org/#/c/234902/, the devstack lib patch needs work, but its seems quite fast now 17:05:10 <dtantsur> Liberty is more or less fine, at least we managed to land all the pending patches (iirc) 17:05:13 <jroll> cool, thanks sambetts 17:05:24 <lucasagomes> sambetts, w00t! 17:05:54 <rloo> dtantsur: so what does that mean wrt kilo -- nothing we can do so no future kilo releases? 17:06:04 <BadCub> morning folks 17:06:19 <dtantsur> rloo, someone has to invest a lot of time in figuring out. nobody did for now :) 17:06:38 <rloo> dtantsur: gotcha. is there a bug against it? 17:06:47 <dtantsur> yep, lemme find 17:07:12 <dtantsur> rloo, https://bugs.launchpad.net/ironic/+bug/1531477 17:07:14 <openstack> Launchpad bug 1531477 in Ironic "Kilo dsvm gate is mostly broken (timeout when waiting for active state)" [High,Confirmed] 17:07:40 <rloo> thx dtantsur, i'll update etherpad if it isn't there 17:08:44 <jroll> anything else? 17:09:09 <jroll> #topic subteam status reports 17:09:19 <jroll> as always, these are here: 17:09:21 <jroll> #link https://etherpad.openstack.org/p/IronicWhiteBoard 17:09:41 <dtantsur> any updates on "Node filter API and claims endpoint"? nothing is written 17:10:52 <jroll> dtantsur: I've kind of put that on hold until the neutron stuff lands 17:10:59 <dtantsur> got it 17:11:16 <jroll> it won't land in nova this cycle, hoping to come back to it in a few weeks 17:11:55 <lucasagomes> jroll, should we hold the indexable json fields too? 17:12:00 * lucasagomes still need to update the spec 17:12:23 <dtantsur> I would try to not hold everything for the next cycle :) 17:12:33 <jroll> lucasagomes: keep working on it 17:12:37 <rloo> we need to get the indexable json fields because that is holding up a lot of stuff like cleaning up how we deal with capabilities 17:12:43 <jroll> I just personally would like to help get the network stuff done faster 17:12:43 <dtantsur> ++ 17:12:58 <lucasagomes> fair enuff 17:13:00 <jroll> and I think doing so is more valuable that a filter api without indexable json 17:13:04 <dtantsur> I was dealing with capabilities for tripleo work: it's terrible :( 17:13:32 <devananda> the neutron stuff is top priority for me right now, fwiw 17:13:52 <jroll> ya, if we let that slip again we've failed hard 17:13:55 <devananda> a lot of people have been putting a lot of work into it for 6 months, and it's very close to done 17:14:35 <devananda> the patch chain works in my local OVS/devstack testing, and so far, the code I've reviewed is (aside from nits) very good -- though I haven't reviewed it all yet 17:14:59 <dtantsur> sounds promising \o/ 17:15:15 <devananda> there's a bit more to do on the CI side - we need to finish the switch to tempest-lib ASAP and rebase a few of the neutron integration patches 17:16:34 <dtantsur> year.. we also need tempest plugin to *finally* finish versioning testing >_< 17:17:36 <sergek> So far I didn't manage to run the tempest plugin ( 17:17:53 <devananda> I think we could land all of the neutron work this week 17:18:05 <jroll> ++ 17:18:20 <devananda> the CI is the main open question for me -- we don't have an experimental job in there yt 17:18:22 <devananda> yet 17:19:11 <jroll> devananda: let's both review the variables in the devstack patches and if they're all good I can push on the project-config patch 17:19:14 <jroll> review them today* 17:19:18 <devananda> getting the tempest lib'ification complete will make testing the neutron integration significnatly easier 17:19:20 <devananda> jroll: agreed 17:19:42 <rloo> wrt tempest lib'ification, what's the delay there? Do we have patches up? 17:20:19 <devananda> vsaienko: do you have time to do the patch rebasing today? 17:21:08 <lucasagomes> rloo, AFAIUI yes, we got the patches creating the base struct and copying the tempest code in ironic 17:21:12 <lucasagomes> and the project config one 17:21:17 <lucasagomes> jroll, anything else missing ? 17:21:52 <jroll> lucasagomes: I have a patch removing our tests from tempest tree 17:21:57 <jroll> and then it's done afaik 17:22:15 <dtantsur> devananda, it's probably pretty late for him.. 17:22:16 <lucasagomes> good stuff 17:22:31 * jroll needs to update that one again for pep8 17:25:26 <rloo> shall we move on? 17:27:32 <mgould> sounds good to me 17:27:56 <lucasagomes> ++ 17:27:58 <TheJulia> ++ 17:28:00 <NobodyCam> :) 17:28:05 <jroll> ya 17:28:09 <jroll> #topic open discussion 17:28:21 <jroll> e0ne: you had something here? 17:28:43 <e0ne> small anouncment from cinder 17:29:10 <e0ne> we've finally got merget "Attach/detach volumes without Nova" spec 17:29:46 <e0ne> and we (cinder team) introduced new cinderclient extension named "python-brick-cinderclient-ext" to local attach/detach features 17:29:53 <e0ne> we discuss it at Summit 17:30:12 <e0ne> it could be used inside Ironic instances to attach cinder volumes 17:30:27 <e0ne> here is or short-term roadmap: http://lists.openstack.org/pipermail/openstack-dev/2016-January/083728.html 17:30:47 <e0ne> we need more feedback and feature requests from you 17:30:55 <devananda> e0ne: woot! 17:31:04 <jroll> \o/ 17:31:19 <e0ne> I hope, we'll merge attach/detach stuff in Mitaka 17:31:33 <TheJulia> Awesome news 17:31:41 <e0ne> Ironic team's iput as very valuable for us 17:31:41 <NobodyCam> nice 17:31:59 <devananda> e0ne: is it possible to test this within devstack today? 17:32:14 <lucasagomes> thanks e0ne! 17:32:22 <e0ne> devananda: yes, I'll share detailed manual with video later this week 17:32:42 <e0ne> devananda: for now, you need just to install this extension from sources 17:33:00 <e0ne> is it's needed, I can update you with our progress on weekly basis 17:33:36 <e0ne> s/is it's needed/if it's needed 17:33:52 <devananda> e0ne: that's fantastic. once there are some docs on how to test / use it, things should accelerate on our end 17:34:15 <e0ne> devananda: I understand it. It's my top priority for now 17:34:53 <e0ne> also, we're thinking about functional tests for it including attach/detach inside ironic instance 17:34:57 <devananda> e0ne: right now, I'm focused on landing the neutron integration, so I won't have time to personally dig into the cinder integration right now 17:35:35 <e0ne> devananda: ok, we've got some time before M-3 milestone 17:35:40 <devananda> anyone else want to take the lead on the cinder work right now? if not, I'll dive into it once the neutron code is landed 17:36:08 <jroll> well 17:36:14 <jroll> let's be clear about cinder; there's two pieces 17:36:26 <jroll> 1) e0ne's work, which shouldn't require ironic changes 17:36:30 <jroll> 2) boot from volume 17:36:33 <lucasagomes> I think there's some people looking at boot from volume but I don't know if there's anyone working on attach/deattach volumes 17:36:42 <e0ne> jroll: you're absolutely right 17:36:47 <devananda> right 17:37:26 <lucasagomes> jroll, +1... 1) will have changes in Ironic when we start attaching/deattaching volumes via the BMC 17:37:39 <lucasagomes> but the brick agent is hardware agnostic 17:37:39 <e0ne> jroll: for the #2 of your list, I need to to take a look on https://blueprints.launchpad.net/ironic/+spec/cinder-integration closer 17:37:50 <jroll> e0ne: ditto 17:38:30 <devananda> lucasagomes: with brick agent, we could start building images that automount volumes on boot 17:38:45 <e0ne> lucasagomes: it's our first step in integration between projects 17:38:50 <devananda> and passing in attachment data via cloudinit 17:38:59 <e0ne> devananda: +1 17:39:06 <lucasagomes> devananda, e0ne sounds good 17:39:23 <lucasagomes> devananda, question there, the volume information is passed via configdrive? 17:39:23 <e0ne> devananda: to be clear, it's not an 'brick agend', it's a cinderclitnd CLI and Python API extension 17:39:25 <devananda> this isn't BMC dependent, and with [i]PXE booting, could also do diskless nodes 17:39:44 <lucasagomes> or it's after the instance is deployed ? (wondering if the provision vs tenant network would be a hurdle here) 17:39:54 <devananda> e0ne: thanks for the clarity 17:40:23 <devananda> lucasagomes: yes, network will be an issue. cinder vol will need to be visible on tenant network for any of this to work 17:40:28 <jroll> lucasagomes: yeah, the instance will need a route to the cinder-volume service 17:40:29 <e0ne> lucasagomes: you can use it after provisioning if you have SSH access 17:40:40 <lucasagomes> gotcha 17:40:47 <e0ne> jroll: actually, co cinder-api and storage networks 17:41:04 <e0ne> I know that ist's not secure enough for multi-tenant envs 17:41:09 <jroll> e0ne: yeah, I usually assume the user can access cinder-api :) 17:41:19 <e0ne> jroll: :) 17:41:37 <jroll> lucasagomes: for instance we have "servicenet" at rackspace - 10. provider network that has an ACL for onmetal -> cinder storage 17:42:15 <jroll> and our cinder backend does magic acl things per volume 17:42:29 <e0ne> jroll: what backends do you use? 17:42:50 <jroll> e0ne: dunno, it's something around LVM 17:42:53 <lucasagomes> jroll, I see, yeah... When time comes we probably should document a reference architeture for those things 17:43:05 <jroll> forget what it's called 17:43:07 <jroll> yeah 17:43:13 <rloo> not to be a spoilsport, but we've spent > 10 min on this... can the rest be taken off line? 17:43:28 <jroll> sure? 17:43:34 <lucasagomes> rloo, yes 17:43:34 <jroll> do we have anything else to talk about? 17:43:37 <lucasagomes> do we have more open topics? 17:43:37 <e0ne> #link http://lists.openstack.org/pipermail/openstack-dev/2016-January/083728.html 17:43:37 <jroll> oh, midcycle 17:44:09 <jroll> so I'd like to do the midcycle in february sometime - does anyone have dates that work great or don't work at all for them? 17:44:18 <e0ne> sorry for the link duplication - now it will be easer to find it 17:44:19 <jroll> or should I just pick one 17:44:42 <lucasagomes> jroll, after fosdem/devconf would be ideal 17:44:44 <dtantsur> jroll, the whole week 1-5th + 15th do not work for me 17:45:09 <e0ne> jroll: we'll got cinder midcycle on January 26-29. ironic itegration questions are on our desk 17:45:23 <jroll> dtantsur: the 15th or the week of the 15th? 17:45:30 <lucasagomes> dtantsur, are you going to the devconf right? 17:45:31 <dtantsur> jroll, no, only 15th itself 17:45:40 <lucasagomes> cause it's 5-7th of feb 17:45:43 <dtantsur> lucasagomes, yes + there're some meeting right before 17:45:44 <jroll> e0ne: I won't be able to make it with that late of notice, I think. link to more details? 17:46:11 <NobodyCam> jroll:this is a virtual midcycle this go round? 17:46:19 <jlvillal> Good chance I will be on vacation for two weeks starting 19-Feb-2015. 17:46:21 <e0ne> jroll: https://etherpad.openstack.org/p/mitaka-cinder-midcycle - we will have online streaming and hangouts call 17:46:33 <jroll> e0ne: cool 17:46:35 <jroll> NobodyCam: yes 17:46:38 <NobodyCam> :) 17:46:59 <e0ne> so it will be great to have someone from ironic there:) 17:47:01 <TheJulia> e0ne: I'm local to the cinder midcycle location and can attend if you guys want a crazy ironic person to attend 17:47:05 <jroll> ok so feb 16-19 looks pretty ideal heh, or 8-12 17:47:24 <e0ne> TheJulia: cool 17:47:40 <lucasagomes> jroll, wfm, maybe propose both dates to the ML 17:47:48 <lucasagomes> to see who would be able to join 17:47:59 <jroll> lucasagomes: yeah, my first priority is cores 17:48:02 <e0ne> TheJulia: it will be helpful for sure 17:48:31 <TheJulia> e0ne: putting it on my calendar then 17:48:38 <NobodyCam> either should work for /me 17:48:44 <e0ne> TheJulia: thanks! 17:48:53 <jlvillal> e0ne, You would be lucky to get TheJulia to attend! :) 17:50:02 <rloo> oh, jroll, maybe mention the new rfc process, or send email about it? 17:50:13 <jroll> did we not send enough emails about it already? 17:50:14 <rloo> err, rfe? 17:50:19 <jroll> all I did was write it down 17:50:24 <rloo> didn't we just update the documentation about it? 17:50:41 <jroll> yeah 17:50:43 <jroll> ok I'll email 17:50:54 <rloo> thx jroll 17:50:58 <NobodyCam> :) 17:51:40 * jroll top posts like a boss 17:51:46 <jroll> anything else? 8 minutes left 17:52:18 <jlvillal> Regarding functional testing and how it gets run. 17:52:24 <jlvillal> There was an email thread started. 17:52:31 <jlvillal> I voted for being able to do it with tox. 17:52:38 <jlvillal> So it is easy to do for developers. 17:52:39 <sergek> yeah. It was me 17:52:46 <jlvillal> There were also options of devstack and tempest. 17:52:50 <jlvillal> Any thoughts? 17:53:02 <jlvillal> I was going to bring it up in the Testing/QA meeting on Wednesday. 17:53:06 <jroll> so 17:53:11 <jroll> those are two different things 17:53:13 <jroll> right? 17:53:22 <dtantsur> inspector does it with tox for developer's ease 17:53:23 <jroll> I haven't seen this thread afaik 17:53:27 <jroll> but like, do it with tox 17:53:40 <jroll> and there's a dependency of having ironic available 17:53:44 <NobodyCam> I would say tox 17:53:46 <NobodyCam> too 17:53:46 <jroll> so if you're in devstack, you have it 17:53:49 <jroll> and can still run tox 17:53:50 <jroll> right? 17:54:00 <jroll> (assuming this is client testing) 17:54:05 <sergek> I think we can run tox with DevStack as well like HEAT does 17:54:09 <dtantsur> jroll, what inspector does (and I think what jlvillal means) is that tox actually starts a limited inspector instance 17:54:13 <jlvillal> Yes. I am thinking for openstack/ironic 17:54:21 * lucasagomes gotta read the ML thread 17:54:29 <dtantsur> jroll, so it's fully standalone, does not need devstack/any other ironic installation 17:54:29 <jroll> oh, ironic, not ironicclient? 17:54:33 <rloo> jroll et al: this thread: http://lists.openstack.org/pipermail/openstack-dev/2016-January/083744.html 17:54:36 <jlvillal> dtantsur, +1 Run ironic-api, ironic-conductor, and rabbitmq. To start 17:54:48 <lucasagomes> sergek, so setup devstack and then call a tox -e<cmd> that will run the tests? 17:54:48 <dtantsur> but I'm just telling what inspector does, not sure I'm fully aware of the context :) 17:55:11 <dtantsur> inspector-client does the same, it uses shared code in ironic_inspector.test.functional 17:55:11 <sergek> lucasagomes: yes 17:55:23 <jroll> jlvillal: that doesn't make it easy for developers, though, it depends on lots of things outside of pip 17:55:25 <jlvillal> I think if we developers have to run devstack and/or tempest to do functional testing, then they likely won't 17:55:34 <sergek> the question was how to run ironic services - manually or with devstack 17:55:39 <dtantsur> https://github.com/openstack/python-ironic-inspector-client/blob/master/ironic_inspector_client/test/functional.py#L172-L174 17:55:40 <lucasagomes> it sounds like a good approach 17:55:42 <jroll> I have to install half of devstack just to start ironic anyway 17:55:57 <lucasagomes> sergek, when we test in gate, we will already have devstack right? 17:56:11 <lucasagomes> this approach sounds good IMO, and it's slim for ironic 17:56:30 <dtantsur> I would love to be able to run 'tox -efunc' on my laptop, if it's even possible.. 17:56:33 * lucasagomes don't want to maintain yet-another big script to setup everything 17:56:36 <sergek> I think we can all options altogether 17:56:37 <jroll> I totally agree 17:56:43 <jroll> but I also don't want rabbit on my laptop 17:56:46 <jroll> is my point here 17:56:57 <dtantsur> jroll, oslo.msg has an in-memory implementation 17:57:00 <dtantsur> iirc 17:57:01 <sergek> *have 17:57:03 <jroll> oh true 17:57:14 <jroll> idk, need to investigate more I guess 17:57:15 <TheJulia> 3 minutes 17:57:17 <TheJulia> s 17:57:33 <jlvillal> I will discuss it more on Wednesday. And feel free to comment in email thread! 17:57:44 <gabriel> zeromq doesn't need broker, might be an option jroll 17:58:09 <gabriel> afaik 17:58:11 <jroll> sure 17:58:14 <jroll> rabbit was one example 17:58:17 <sergek> I'd love to have all three options - manual service, devstack and tempest plguin 17:58:31 <sergek> alas I didn't manage to start Tempest plugin yet 17:58:37 <dtantsur> sergek, I don't see value in manual service option tbh.. 17:58:44 <sambetts> vagrant ? 17:58:48 <dtantsur> I'd like to call tox -efunc and just have it finish to the end.. 17:59:07 * dtantsur sets a reminder to respond to the thread tomorrow morning 17:59:09 <jlvillal> +1 on tox -efunc 17:59:17 <sergek> dtantsur: agree. and the question is what happens under the hood 17:59:17 <jroll> ok we're out of time 17:59:23 <sambetts> https://www.vagrantup.com 17:59:24 <jroll> #endmeeting