16:02:42 #startmeeting ironic_neutron 16:02:43 Meeting started Mon Aug 31 16:02:42 2015 UTC and is due to finish in 60 minutes. The chair is Sukhdev. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:02:44 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:02:46 The meeting name has been set to 'ironic_neutron' 16:03:00 we are in business :-) 16:03:11 #topic: Agenda 16:03:25 #link: https://wiki.openstack.org/wiki/Meetings/Ironic-neutron 16:03:45 Anybody wants to add anything to the agenda? 16:04:17 #topic: Announcements 16:04:29 Neutron FF is later this week 16:05:00 We are good from the Neutron FF point of view except for ML2 documentation 16:05:25 There is Neutron core meeting later today, I will double check if we are bound by the deadline for that 16:05:53 This FF is not applicable to vendor repo's for ML2 drivers 16:06:06 docs are never blocked by feature freeze iirc 16:06:17 or from what I've seen, anyway 16:06:37 jroll: right, but, still it is good to double check 16:06:51 Any body has any other announcement? 16:07:15 if not, lets dive into the agenda 16:07:35 #topic: Plan actions from last week 16:07:55 We covered a lot last week in terms of the plan for the week 16:08:19 1) I was suppose to follow up with amotoki regarding documentation - 16:08:42 I made several attempts - have not been able to - he had family emergency - 16:09:10 will continue to do so, otherwise, I will write up the document myself 16:09:38 lazy_prince: was suppose to upload the patch with authentication issue resolved 16:09:58 I saw updated patch - have not had time to review it or test it 16:10:17 lazy_prince: does it address the admin context issue - the one we discussed with devananda? 16:10:44 yea.. so i have the patch but need to resolve some merge conflicts while rebasing.. will be done by tomorrow.. 16:11:04 Hopefully. it should unblock you.. 16:11:14 lazy_prince: the fix suggested by devananda worked? 16:11:29 in my env, it worked.. 16:11:46 yeah, get_admin_context() is a thing that works :P 16:12:07 lazy_prince: perfect - when you are able to resolve the rebase issue, ping me, I will get it tested 16:12:16 sure.. 16:12:24 jroll: we missed you last week :-) 16:12:31 Sukhdev: :) 16:13:17 cragusa: You were going to work with Will to push the updated patch for the DB issue 16:13:29 cragusa: any luck with that? 16:13:36 Will fixed that 16:13:53 and pushed the patches already 16:14:04 AFAIK should be working 16:14:23 cragusa: cool - I must have missed the updated patch - 16:14:36 was done earlier on today 16:14:45 Sukhdev: can you test it? 16:15:01 chigang: ah that makes sense 16:15:32 chigang: yes, I will look for it and test it along with lazy_prince's patch - both are needed to get us going to the next step 16:15:45 chigang, lazy_prince: thanks for the updates 16:16:11 #topic: Integration Effort 16:16:19 We covered this topic last week - 16:16:54 We were blocked (for jroll's benefit - to catch up) in our integration effort because of couple of issues 16:17:35 now that we have updated patches, I will go back and test it and see if this unblocks us to move forward with the integration testing 16:18:21 any body wants to add anything to what I have already said? 16:18:39 well, moving right along 16:19:09 #topic: Discussion on the planning for next few weeks until the release 16:19:42 Now that we have all players present here, I think it will be a good idea to discuss the plan for next few weeks 16:20:47 I think we should distribute the work and all pitch in to test various parts to make sure we are in good shape to be able to get all Ironic patches merged in time 16:21:42 One of the issues - CI testing 16:21:44 I still see gate-ironic-python27 failing for the patches.. 16:23:05 yes - we need to ensure that basic gate issues are in order - so that once the code gets reviewed, we are blocked from merges 16:23:16 lazy_prince: are you taking over all of the patches? 16:23:21 jroll had put some comments on the etherpad about it 16:23:25 or is everyone just doing it? 16:23:45 oh, right 16:24:09 so network_provider defaulting to neutron shouldn't make patches fail CI until it's used 16:24:17 but even the first patch in the chain is failing 16:25:13 that was what i wanted to come to.. :) first patch failure.. 16:25:24 * yhvh was in the wrong room :( 16:25:38 yhvh :-) 16:25:49 and for the network_provider, i have made it as default to none... 16:26:08 lazy_prince: this looks like the new WSME release, I think, rebase might just fix it or you could poke lucas about it 16:26:20 for time being and then will switch to neutron once all other patches are working.. 16:26:25 lazy_prince: something about throwing 400 for not-exposed fields in req/resp object 16:26:59 probably yhvh should take a look at it.. 16:27:12 are these the current test failures? 16:27:37 yes 16:27:39 http://logs.openstack.org/32/206232/21/check/gate-ironic-python27/4a76b62/console.html 16:27:40 I will look into it 16:27:43 ty 16:27:57 thanks! 16:28:38 yhvh: py27 errors on https://review.openstack.org/#/c/206232/ - as an example 16:29:37 lazy_prince yhvh: are there any issues that are blocking you? 16:30:04 no, but I did want to ask Om to carefully review the rebase I did on his patch today 16:30:09 #action: yhvh to look into py27 related errors in the gate 16:30:11 nope.. I just have to finish the rebase... 16:30:40 yhvh: I assigned you the action - hope you are OK with it? 16:30:47 yes 16:32:01 So, I am concentrating on testing two pieces - 1) to make sure when port groups of switch-ids are configured, they are correctly populated into the DB 16:32:29 2) and to verify that the correct switch-id, (profile info) correctly flows to the ML2 drivers 16:32:37 my recent patchsets should help with 1 16:33:17 yhvh: there was some documentation issue (i.e help strings issue with the CLI) - hope you fixed it as well 16:33:38 Other than 1 and 2, there are more pieces need to be tested 16:33:56 Sukhdev: not pushed this yet, just the db bug 16:33:58 We need volunteers to test other parts as well 16:34:33 yhvh: last time we had to struggle a bit to figure out the format for the data-entry for the CLIs 16:35:13 yes fyi it's like: local_link_connection='{"some":"value"}' 16:35:21 yhvh: hopefully, your next update will fix this 16:35:41 yhvh: yup - we figured it out :-) 16:36:10 yhvh: thanks 16:36:39 So, going back to my previous question - we need to test the other parts as well - 16:37:07 how do we plan that testing? 16:37:29 to make sure that all pieces are going land in place 16:38:23 any ideas, thoughts? 16:38:55 we are still setting up our CI infrastructure, we just hired a CI guy so things should move faster 16:39:07 but we can't promise anything at this time 16:39:52 \what other parts need to be tested? 16:40:38 jroll: all negative test paths - ensure we basically rugged - 16:41:17 jroll: I am only testing the positive path - i.e. I can configure the profile information - and make sure it gets to the ML2 driver 16:41:37 ah, I see 16:42:04 I am not testing any negative test cases - e.g. if I mis-configre things - those are detected correctly 16:42:24 or I configure the same interface over and over - what happens, etc..etc.. 16:42:29 you get the idea... 16:43:17 Another thing I am trying to test is the network flip - 16:43:45 but, it is not ready yet - because of blocking issue - hopefully I can progress this week on this 16:44:06 so on that... I have a question.. if a network does not have any subnet associated, should the port creation be allowed..? 16:44:25 I noticed one issue - (it is listed on ehterpad) on the first create_port(), we need to ensure that vnic_type is set correctly - presently it is not set 16:44:58 jroll: perhaps this fix can go with your patch - when the host_id is set, this can be set as well 16:45:28 Sukhdev: yeah, I'll try to update that soon 16:45:39 lazy_prince: can a neutron network exist without a subnet? 16:46:04 yes.. just net-create and no subnet-create.. 16:46:19 * j1mehta slaps blogan around a bit with a large fishbot 16:46:24 jroll: it can - but, we need it in order to create a port on it - to get appropriate IP address 16:46:48 sigh 16:46:56 that was interesting 16:47:22 lol 16:47:30 blogan:: was it interesting or painful - :-) 16:47:31 so basically such call should fail... 16:48:03 lazy_prince: yes, if you do a create_port without subnet, I think it will fail 16:49:08 Sukhdev: both? i feel violated 16:49:26 blogan: :-) 16:49:27 i was thinking, it could be useful for some cases where you just put the port on network and then use the port for ovs switch etc... but anyways.. we will see that later.. 16:51:25 lazy_prince: did not understand it - so, you want to create a subnet later, after having created ports on a network? How do you assign the IP address? 16:52:05 * Sukhdev time check - only 9 min. left 16:52:17 lets talk about it later on... for now lets go with the flow on the topic at hand.. 16:52:28 lazy_prince: +1 16:53:17 Also, please make sure to update ehterpad to list all correct patches - I am only using the etherpad 16:53:53 there are few patches there that should be abandoned - and notice some patches floating around that are not on the etherpad 16:54:46 can everybody take time to review it and make sure your patches are listed there - and anything that does not belong there (ie. irrelevant or old) remove it from there 16:54:59 sure 16:55:13 remove or strike-through... 16:55:17 we will use etherpad to track all the code and track the merge status 16:55:41 lazy_prince: either one is OK - strikethrough may be better 16:56:01 yeah.. that will preserve history though.. 16:56:22 #topic: Open Discussion 16:56:29 we are down to 4 min. 16:56:41 anything we have not covered, needs discussion? 16:57:20 we have some patches for ironic-inspector and diskimage-builder to add the extra info we need during the discovery phase, is this interesting to you? 16:57:26 jroll: when you update your patch, please ping me - I will include those in the testing 16:57:36 Sukhdev: ok 16:57:57 yhvh: that is not in the scope for this phase 16:58:01 yhvh: yes, those are interesting, please put them up :) 16:58:13 * yhvh flips a coin :) 16:58:17 they're lower priority but we can do that work in parallel 16:58:36 ok I was holding off to keep attention on current work 16:58:51 jroll: I though we did not want to do the discovery part as a part of this effort 16:59:14 Sukhdev: no, but if someone else wants to then we should do it 16:59:25 Sukhdev: the inspector and DIB review teams are different teams, too 16:59:32 so not much resource constraint 16:59:54 jroll: as long as it does not impact our work, I am OK 17:00:08 +1 17:00:14 ok I will post this week 17:00:15 it doesn't do much without our work getting done anyway :) 17:00:39 yhvh: create a separate section on the etherpad and you can list it there - 17:00:55 Sukhdev: kk 17:00:58 we will not address them here 17:01:06 We are out of time - 17:01:12 thanks everybody 17:01:16 bye 17:01:19 #endmeeting