19:00:04 #startmeeting Ironic 19:00:04 #chair devananda 19:00:04 As always agenda for the meeting is at: 19:00:04 #link https://wiki.openstack.org/wiki/Meetings/Ironic 19:00:04 Meeting started Mon Sep 23 19:00:04 2013 UTC and is due to finish in 60 minutes. The chair is NobodyCam. Information about MeetBot at http://wiki.debian.org/MeetBot. 19:00:05 #topic Greeting and roll-call and announcements. 19:00:06 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 19:00:08 The meeting name has been set to 'ironic' 19:00:09 Current chairs: NobodyCam devananda 19:00:21 whos here for the ironic meeting? 19:00:26 \o 19:00:28 o/ 19:00:37 * lucasagomes very sleepy but here 19:00:42 lol 19:01:03 lucasagomes: if you were still on US time, it'd barely be lunch time now :p 19:01:16 Great to meet many of you at the seattle meetup 19:01:29 should have changed that to all but 19:01:30 devananda, haha that's the prob, I think I need at least one more day to get back on track 19:01:50 Quick wrap-up from last weeks meeting: Yes romcheg and I had a good meeting last week :-p 19:01:53 I had a action item: Ask devananda to think about a Transifex project for Ironic 19:01:56 devananda any Thoughts / comments on ^^^^^ 19:02:04 yes 19:02:36 I pinged clarkb about that last week and created a transifex account. AIUI, it's just a matter of him creating something for us 19:02:50 and then a small patch to openstack-infra/config 19:03:15 clarkb: or if I forgot something, pls let me know 19:03:19 w00t 19:03:44 any body else ( lucasagomes ) 19:03:45 lol 19:03:52 good stuff :) once it's setup I'll translate ironic to brazilian portuguese as well 19:03:54 * NobodyCam loves small meetings 19:04:01 also, the transifex integration will auto-generate the .pot files for us (part of what it does). 19:04:15 so for now we should continue to -- patches which accidentally change that file 19:04:23 :) 19:04:53 can that be blocked automatically? 19:04:54 #action Reviewers will -1 patches that touch the .pot file 19:05:05 can we remove this file 19:05:29 let me check other projects real quick. we might be able to 19:06:33 while deva does that 19:06:35 Hong Kong design summit: `Your Papers Please!` 19:06:35 We have three! Submit yours here: 19:06:36 #link http://summit.openstack.org/ 19:07:05 NobodyCam, +1 yea will do 19:07:12 :) 19:08:14 Hi NobodyCam, I am trying to figure our what belong to Ironic and what to Tuskar, 19:08:18 should we move on? and come back to the .pot file 19:08:23 I linggao 19:08:32 For example, hardware discovery. 19:08:35 doesn't look like it's in the .gitignore of other projects at all 19:09:05 linggao: we do have a story for that 19:09:16 ? 19:09:18 linggao: there's a section for open questions at the end of the meeting. 19:09:26 that may be best handled not in this meeting 19:09:54 or there too! :) 19:09:57 oh, I thought you are talking about sumit papers. 19:10:15 I posted that while deva was looking up stuff 19:10:17 linggao: oh! gotcha. I thought you meant code wise :) 19:10:31 linggao, dicovery is Ironic... tuskar is more about modeling ur deployment using a high abstraction 19:10:54 we can for sure talk about papers 19:10:55 linggao: if you have a specific topic in mind, please mention it either here or on the ML, and we can help figure out which track it is best suited for 19:10:58 and also they will offer a nice ui (extension to horizon) to deploy baremetal 19:11:40 yes, I was thinking about automatically discover the bearemetal nodes, and put them in the database. 19:11:49 linggao: hardware discovery falls into both ironic and tuskar but in different ways 19:12:23 linggao: ironic will need to provide some API for hardware interrogation, eg. via IPMI or a bootable ramdisk, but ironic stops there. it's _just_ the API. 19:12:47 linggao: tuskar (or something else?) could then model that data and do something with it (eg, display it, or make scheduling decisions, etc) 19:12:51 does that help? 19:13:06 sounds correct to me 19:13:09 :) 19:13:27 oh the web we weave 19:13:34 thanks devananda for the clarification. 19:13:39 go to move on? 19:14:02 y for me 19:14:12 #topic Testing 19:14:12 ++ 19:14:25 h thats me. 19:14:25 TripleO is making changes to the heat templates so some of the patches will 19:14:28 undergo changes to conform to the changes. I will stay on top of the current patches. 19:14:31 We do howerver have a working walk through that will install Ironic and the Ironic cli client 19:14:34 on the tripleO undercloud image. see 19:14:36 #link https://wiki.openstack.org/wiki/Ironic#Using_Disk_Image_Builder 19:14:39 on how to info. 19:14:53 has anyone looked over the new walk thru 19:15:07 I got up to starting the undercloud VM on saturday 19:15:17 but ran into problems with missing ENV vars taht weren't in either doc 19:15:31 ?? which ones 19:15:38 not in our wiki, and not tripleo/devtest 19:15:50 which env var 19:16:07 * devananda restarts seed VM ... 19:16:13 do you recall 19:16:14 these VMs don't like to hibernate 19:16:17 lol 19:16:18 no 19:16:36 I have a question on testing. 19:16:39 I will run thru again today 19:16:41 sure 19:16:47 shoot linggao 19:16:55 in order to make things easier should we use a seed vm to bootup our undercloud? wouldn't be easy to create an undercloud straight away? 19:17:09 undercloud vm* 19:17:18 one of the veiw I got says that I should use mock instead of mox because openstack is moving to mox. 19:17:18 thats a tuff question 19:17:21 yes and not 19:17:25 *no 19:17:28 can anyone expalin this? 19:17:33 linggao: sure 19:17:44 the undercloud is the first real cloud 19:17:58 linggao: mox is not python3 compatible, and there is a directive for all of openstack to move towards py3 compatibility 19:18:20 everything in seed is faked 19:18:41 linggao: right now, ironic's unit test suite is a mixture of mox (inherited from Nova and other projects) and mock (new code). 19:18:56 NobodyCam, right... but still, I mean for testing we don't really need to launch a seed vm, we can just start with a undercloud in a virtual machine and from that we can deploy the overcloud 19:18:57 linggao: we're moving towards all the unit tests being done with mock 19:19:11 it would save some time + memory 19:19:22 lucasagomes: except "undercloud" requires a working cloud init, and it is configured via Heat in the seed cloud 19:19:38 devananda: ++ 19:19:42 devananda, I see. I will change it to mock then. It is a lot of changes. ouch. 19:19:43 lucasagomes: what you're really proposing is taht we merge ironic into the seed VM 19:20:23 devananda, kinda yea... I mean we still can fake the cloud init thing adding a json on the /var/.../heat-cfn directory no? 19:20:27 we can do that 19:21:11 ironic currently can not actually deploy thou 19:22:10 votes for working ironic into seed? 19:22:56 well I'd go for it at least while ironic can't deploy an image 19:23:12 im actually gong to vote -1 at least untill we have a working ironic 19:23:22 just to make it easier for people to get ironic-api/conductor/database etc working and develop on it 19:23:24 linggao: I'd like to see native-ipmi landed before the summit so folks can start playing with it then 19:23:40 linggao: if moving it to mock is going to prevent that, please let me know 19:23:40 I would vote, I think the ultimate issue would be simplicity 19:24:01 lucasagomes: I use seed to create and distroy my test undercloud 19:24:12 i think that actualy saves time 19:24:13 NobodyCam: I would still prefer ironic in the undercloud for now so we can do AB testing with nova-baremetal 19:24:21 lucasagomes: ^ 19:24:28 I just load the new image into glance anfire away 19:24:52 my plan for testing deployments of ironic is to use the undercloud layer, and run two underclouds -- one nova-bm and one ironic 19:25:04 so that I can do A/B testing and develop a migration script 19:25:16 devananda, I need to learn mock. I'd be really happy if we can let it go throgh and I'll make next feature mock compatable. 19:25:17 right... yea well we still have devtest if all u want is to setup a quicker dev environment 19:25:36 lucasagomes: you mean devstack? 19:25:46 devananda, yes sorry 19:25:50 devstack 19:25:54 thats there 19:25:55 :) 19:25:59 :) 19:26:30 * lucasagomes slaps his own face :P 19:26:35 NobodyCam: thoughts on accepting the native-ipmi driver with mox for now, so the code lands and can be worked on, then migrating it to mock for py3 compat later? 19:27:06 devananda: sure we there is a TODO to replace with mock 19:27:11 I'm fine with this ^, fwiw. add'l tech debt is not ideal, but keeping functionality out of trunk right now is not good for speed 19:27:24 cool 19:27:50 :) 19:27:56 thanks everyone :) 19:28:02 ty linggao :) 19:28:09 linggao: I'll open a bug about mox->mock for ipmi-natie and assign to you :) 19:28:18 any thing else on testing? 19:28:30 nope 19:28:32 maybe we can suggest openstack-infra also to only accept libs that are compatible with py3 from now on 19:28:40 devananda, sure, I like bugs :) 19:28:49 there is a py3 gate job 19:28:53 or maybe they r already doing that, idk 19:29:11 o moving on... 19:29:13 #topic in-progress tasks 19:29:14 41976 Abandoned??? 19:29:47 https://review.openstack.org/#/c/41976 19:29:58 no yuriy today 19:30:04 yea :/ 19:30:32 should be be looking at picking up on that patch? 19:31:18 I don't want to step on any toes but we do kninda need that... don't we? 19:31:33 hmm let's see if he will reopen it 19:31:36 we can ping him tomorrow 19:31:40 ya 19:31:44 and see, if we get no answers 19:31:53 then we do it 19:32:03 In the f2f meetings last week, I seem to have overlooked the discussion that happened on that patch 19:32:34 thats just it nothing has happened 19:32:53 NobodyCam: he proposed changing the signature of VendorInterface.validate() to return (bool, bool) 19:32:56 instead of just (bool) 19:33:23 I agree with Lucas' -1 on that 19:33:40 yea IMO should use an exception there 19:33:43 like here 19:33:49 https://review.openstack.org/#/c/41976/8/ironic/conductor/rpcapi.py 19:33:59 ya.. we were going to talk about that and just havent had the chance to hook up 19:34:15 instead of having this async_continue variable there, we should just except one exception and then it means that the validation failed 19:34:22 ValidationError() 19:34:33 lucasagomes: +1 19:34:37 lucasagomes: i think the point is, how to determine whether to make a cast() after the call returns 19:35:14 devananda, if validation pass do a cast, if it fails (exception raised) do not do it 19:35:18 if validation passes then make the cast 19:35:27 lucasagomes: :-p 19:35:45 except someetimes, you only want to call 19:35:50 even when it doesn't raise an exception 19:36:16 let's pick that one up after the meeting :) 19:36:20 only = dont ??? 19:36:57 we really at fft so. 19:37:00 #topic Food For Thought / Open Discussion 19:37:05 #action devananda to dive into review 41976 (pass_vendor_info) 19:37:40 already fft? 19:37:41 o wow 19:37:46 NobodyCam: we skipped the API discussion -- and I think there's some things to cover there 19:37:57 how about a wrap up of the f2f devananda and lucasagomes had on friday 19:38:06 gah I did 19:38:32 Do you still have Ironic meeting? 19:38:40 romcheg_, welcome :) 19:38:41 yes 19:38:42 romcheg_: welcome :) 19:38:46 heh 19:38:54 Hi then :) 19:38:54 #topic API discussion 19:39:03 welcome romcheg_ 19:39:12 #link https://bugs.launchpad.net/ironic/+bug/1227431 19:39:13 sorry, got asleep 19:39:14 Launchpad bug 1227431 in ironic "API request to list resources should return only minimal data" [High,In progress] 19:39:31 lucasagomes: you beat me to it - i was just about to paste that :p 19:39:36 #link https://bugs.launchpad.net/ironic/+bug/1227928 19:39:37 Launchpad bug 1227928 in ironic "API collection not named based on resource type" [High,Fix committed] 19:39:47 devananda, :D 19:40:02 so that's two bugs we opened on Friday regarding the API 19:40:14 the 1227928 is already fixed/merged 19:40:24 anyway, I'm on my smartphone today, the laptop died, so ask questions, if you have any 19:40:26 right, so lucas and I talked on friday about some API issues which, btw, became really easy to spot now that NobodyCam's dib/heat templates are working and I have a running ironic service :) 19:40:31 I'm working on the 1227431, I hope I will fix that tomorrow 19:41:02 romcheg_: lucasagomes and devananda found some instrstering bugs on friday 19:41:34 NobodyCam: will take a look at them 19:41:49 romcheg_: tl;dr; from the first 1/2 of meeting -- transifex stuff is in progress, waiting on clarkb for a few changes in infra/config 19:42:01 devananda, another thing about the 1227431... the "minimal data" also returns links to the resource itself: http://api.openstack.org/api-ref-compute.html 19:42:26 NobodyCam: that's great! 19:42:47 looking forward to start translating 19:42:57 romcheg_: links are in the scroll back and also on the agenda page too 19:42:57 lucasagomes: ah! so it does 19:43:10 yup :) 19:43:44 the links for the sub resources will only be retrieved with the /details 19:44:20 lucasagomes: right. or to /nodes/{uuid} 19:44:27 yup yea 19:44:39 sounds good 19:44:40 nodes/{uuid} always returns the full document 19:44:52 shall we move on to fft? 19:44:57 :) 19:44:58 yup 19:45:01 #topic Food For Thought / Open Discussion 19:45:08 fft 19:45:26 I will on and off line of the next several day moving to cal 19:45:28 I almost finished my work with tempest 19:45:36 devananda, NobodyCam: that's the deploy ramdisk you guys talked about on seattle? https://github.com/theforeman/foreman_discovery/ 19:45:39 #link https://github.com/theforeman/foreman_discovery/ 19:46:04 discovery ramdisk* 19:46:07 romcheg_: how is tempest testing ironic at this point? (it can't do a full deploy yet...) 19:46:35 lucasagomes: that looks like ruby?? 19:46:35 devananda: now I only did some API tests 19:46:46 also wanted to thank lucasagomes for the working ironic command line cli!!! 19:46:50 go show chassis 19:46:58 devananda, yea it's ruby 19:47:11 so might not be it hehe 19:47:13 that's a good question, how should we test deployment 19:47:20 romcheg_: ah! framework is great at this stage -- we'll add more as more is possible :) 19:47:32 we need some hardware for that 19:47:46 NobodyCam, :) will add more things to the cli/lib once I get the api more established 19:47:49 romcheg_: there are some folks already working on testing tripleo deployment, both in VM and on hardware. you should talk with them 19:47:51 some hardware on Infra's side 19:48:14 also going to add a top level section to the meeting for client stuff 19:48:45 romcheg_: IIRC, it's pleia2 and dprince working on TOCI and the -infra side of tripleo 19:49:07 thanks devananda 19:49:21 and my last tid bit : 19:49:28 #link https://github.com/NoBodyCam/ironic/blob/master/README.rst 19:49:55 its our Java driver 19:49:57 lucasagomes: foreman is definitely _not_ what I was referring to w.r.t. a ramdisk agent 19:49:58 lol 19:50:09 devananda, :D 19:50:17 do you have the link to the right one? 19:50:27 lucasagomes: nope. haven't found it but still looking 19:50:33 here's a quick brain-storm on other FFT ideas.... 19:50:44 I just did a small search about it and bumped into that one, so I thought: "well maybe" 19:50:48 lucasagomes: the link we were looking for was an example 19:50:55 nova-baremetal -> ironic: incrememtal code migration? No. New driver? Yes. ++ a migration script. 19:51:12 we will need to create it? 19:51:35 Neutron patches for dynamic DHCP ports -- will require some changes to our PXE driver and start-up scripts (eg, no more local dnsmasq). 19:51:52 w00t 19:51:54 aim to do ^ during I-1 or no later than I-2 19:52:10 ghe seems to be on holiday 19:52:13 lol 19:52:18 models instead of migrations? I don't understand why we'd do taht when all other projects are using migrations 19:52:52 inform node of ironic API URL? yea. those all look reasonable :) 19:52:56 that was the general feeling... I'll remove that one 19:53:00 I think jog0 had a good argument to keep migrations ^ 19:53:09 yes 19:53:28 more than 1 conductor? yep. that's part of our HA story and one reason we broke out of Nova 19:53:42 heheheh 19:53:42 it's probably worth a design session jsut for the more-than-one-conductor story 19:53:53 oh you mean: 19:53:55 Hong Kong design summit: `Your Papers Please!` 19:53:55 We have three! Submit yours here: 19:53:55 #link http://summit.openstack.org/ 19:53:58 lol 19:54:06 #action devananda to propose dev session for more-than-one-conductor 19:54:09 :) 19:54:21 w00t that would be 4 19:54:37 :D 19:55:10 a 19:55:10 [end brain storm] 19:55:15 any thing else? 19:55:30 should we let romcheg_ and lucasagomes get some sleep 19:55:33 lol 19:55:40 +1 19:55:41 haha 19:55:43 :) 19:55:51 thank you all great meeting 19:55:59 #endmeeting