20:00:11 #startmeeting Octavia 20:00:12 Meeting started Wed Nov 30 20:00:11 2016 UTC and is due to finish in 60 minutes. The chair is johnsom. Information about MeetBot at http://wiki.debian.org/MeetBot. 20:00:13 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 20:00:15 The meeting name has been set to 'octavia' 20:00:45 I know there are more people around than just the two of us.... 20:00:58 #topic Announcements 20:01:09 Octavia has moved out of the neutron stadium to the OpenStack big-tent! 20:01:09 howdy 20:01:17 yay yay yay 20:01:17 #link http://governance.openstack.org/reference/projects/index.html 20:01:20 beers on johnsom :D 20:01:23 o/ 20:01:29 congrats mr. new ptl! 20:01:39 In case you have not heard, octavia is now a top-level OpenStack project. 20:01:49 (officially) 20:02:14 So, congratulations are in order for the whole team that has helped us get the project to this point. 20:02:44 congratulations! 20:02:46 :) 20:03:01 Now it means we really have work to do! Grin. 20:03:43 We have come a long way in just a few years, so, yes, raise a glass of your favorite beverage tonight to work well done. 20:04:18 Also, reminder the PTG registration is open 20:04:29 #link http://www.openstack.org/ptg 20:04:42 Hi 20:04:43 does the new governance affect this ptg? 20:05:01 If you are registering, we will be sharing space with neutron, so please register for the neutron room. 20:05:35 dougwig Yes, they offered to give us a room, but given the low interest from the team I opted to continue the plan to share with neutron. 20:05:50 johnsom: ack, thanks. 20:05:54 Plus we still have some coordination to do with them, i.e. client planning 20:06:31 sounds good 20:06:38 Ocata milestone 1 has passed, Ocata-2 is week of December 12th 20:06:45 #link https://releases.openstack.org/ocata/schedule.html 20:07:15 Ocata is counting down, so please focus as much as you can on our Ocata goals. 20:08:06 nmagnezi Please review the timeline so we can get the patches you want into Ocata. 20:08:34 Any other announcements today? 20:08:35 johnsom, my lbaas patches are still up for review 20:08:41 I can link them 20:08:56 #link https://review.openstack.org/#/c/344658/ 20:09:04 Please do 20:09:06 #link https://review.openstack.org/#/c/299998/ 20:09:26 the latter I need to fix couple of tests who broke after a rebase, but nothing major 20:09:48 the first one is ready for your reviews 20:09:55 Ok, on that note: 20:09:56 nmagnezi: 29 is failinig on jenkins 20:09:57 #topic Brief progress reports / bugs needing review 20:10:21 I wanted to ask about removing flask library. I want to remove it from distributor as well. Could u show me some commit doing it with a phone. I didn't find such 20:10:22 diltram, yes sir, I will fix it soon :) broke due to a rebase 20:10:44 nmagnezi: ok, great 20:10:54 With amphora 20:10:56 perelman https://review.openstack.org/386758 20:11:07 Thanks, 20:11:16 diltram, the other patch passed Jenkins so you may review it (please) now 20:11:36 :q 20:11:44 sorry 20:11:46 ok, I'm a core there because of becoming top-level so I will take a look into that 20:11:58 I have been continuing to work on the quotas patch between working on the governance change, reviews, gate fixes, and some testing changes. 20:12:52 Currently I am working through some ORM fun caused by the quota check transactions. It's close to done though. 20:13:50 I know diltram got the policy patch done and it is up for review now too. 20:14:01 #link https://review.openstack.org/399117 20:14:14 in agenda I added a links to review 20:14:28 Yeah, second part of keystone: 20:14:31 #link https://review.openstack.org/#/c/265065/ 20:14:41 and the build rate limit 20:14:48 #link https://review.openstack.org/#/c/303304/ 20:15:36 Please review patches, even if you are not a core reviewer. It helps speed up the process 20:15:37 plus I'm working right now on pagination and sortingg 20:16:08 I'm rewriting that code to use pecan hook 20:16:15 After I am done with quota I plan to get back to the api reference docs 20:16:22 to make it as simple as possible to all controllers 20:16:53 Also, you might note, the start of the /v2 API work for octavia has been posted, so work is starting there. 20:17:41 Any other progress reports? 20:17:44 ankur-gupta-f and sindhu are working on moving types and controllers from v1 to v2 20:18:09 #topic Open Discussion 20:18:09 so any questions to them :) 20:18:31 johnsom: we are currently distracted with comparing the Responses from Octavia to what is expected. 20:19:00 i.e. Octavia has the type 'enabled' the equivalent would be 'admin_state_up' is what Neutron is expecting 20:19:10 how would one go about making the translation 20:19:28 ankur-gupta-f Yeah, there are going to be a few items like that 20:19:52 They are going to need to be translated into the format neutron-lbaas v2 responds with 20:20:15 and accepts 20:20:20 correct. We have the JSON diff comparing the responses and have been able to identify what needs to change. Just need some insight into how to go about changing 20:20:59 http://pasteboard.co/4hRNzLxkR.png 20:21:23 ankur-gupta-f Do you want diltram or I to work with you or are you still working on it? 20:21:36 either of your two 20:21:43 input would be greatly appreciated 20:21:47 atleast to get us on the right track 20:21:50 Ok, yeah, NP. 20:21:56 we can take it offline 20:22:13 so for now I'm working on this pagination 20:22:31 I don't really want to take more on my head right now 20:22:40 BTW, I see that project_id and tenant_id are offset. My plan on that was to support both in the v2 API. tenant_id should be deprecated, but still work in our new API 20:22:48 if you need help just ping me 20:23:24 ankur-gupta-f I will ping you after the meeting and we can talk 20:23:29 perfect 20:24:10 blogan dougwig Has progress been made on the vendor driver code? 20:25:09 Crickets 20:25:21 I need to follow up on that. 20:25:25 sec 20:25:33 blogan came right now ;P 20:25:40 johnsom: hi 20:25:49 johnsom: i have yet to get with dougwig, he's a hard fellow to catch 20:26:14 but i've done a bit of work on the driver stuff, basically just refreshing myself on it and fixing some minor things 20:26:28 blogan I'm getting a bit worried on timing/progress 20:26:57 yes, so am i. to thoroughly test it i will need dougwig's setup 20:27:06 the a10 stuff he setup is no longer active 20:27:40 blogan Would an alternate driver work? I may be able to find another vendor willing to help out 20:28:05 how about our friends from F5? 20:28:22 johnsom: i may see if i can use the namespace driver...but every driver is going to be a bit different and as such some things will have to be added jsut for those drivers 20:28:38 blogan namespace is a good plan 20:28:42 but the idea is that it should be relatively simple to add these things 20:29:25 I know netscaler used to have a free appliance you could download. It was capped for performance, etc. but would be perfect for this. 20:29:30 johnsom: well the namespace is going to be a bit different bc it uses teh agent infrastructure that neutron sets up...which no other driver uses and its not manipulatable (is taht word?) via API 20:29:43 Let me know if you want me to shake the trees and get you some code 20:29:51 aight 20:30:56 Also, radware and vmware are pretty active, I'm sure we could ask those folks for assistance as well. 20:31:19 Ok, any other open discussion items? 20:31:22 oh, i step out for a coke and it's other vendors? :) 20:31:27 blogan: time today? 20:31:37 dougwig Grin... 20:31:52 dougwig: i can make time for you buckaroo 20:32:11 I like it. Let me know if I can help as well 20:32:32 blogan: so, step one is getting you a local setup, i think 20:32:33 Any questions about the governance change? 20:33:40 sorry all.. trying to do too many things in parallel (and not paying close attention here). I saw an F5 reference but missed the context. 20:34:12 we need some hardware to test drivers api in octavia 20:34:27 blogan is moving it from neutron lbaas into octavia 20:34:51 what manner of hardware? plain servers, or vendor-specific (F5 chassis)? 20:35:04 i would wager that virtual appliances would be more useful, actually. 20:35:04 m-greene We were talking about vendors that could provide a test setup to be used when migrating the vendor driver shims over to run under Octavia. (doesn't have to be HW I suspect) 20:35:09 diltram, m-greene: i'm nto exactly moving it, just making a move simpler, hopefully 20:35:28 blogan: can i send you a qcow2, or ova, or what? 20:35:39 dougwig: send me a qcow2 20:35:55 yeah VM will be easier to test probably 20:36:11 blogan: sign up here, and i'll make it free for you: https://glm.a10networks.com/wizard/glm_welcome/create_account 20:36:55 Nice, thanks dougwig. I really wanted to make sure we were moving forward on the driver part. 20:37:19 if we can get blogan a lab with multiple vendors, that'd be awesome. 20:38:05 preferrably a lab where i can call myself a evil genius 20:38:09 I will help get him access to the netscaler stuff. I've used it in a previous life for this kind of work. 20:38:19 johnsom: back when you were a cat? 20:38:43 * johnsom Thinks blogan means a lab with a big fat reset button when he breaks everything 20:38:57 dougwig: created account with brandon.logan@rackspace.com 20:39:34 This brings up the topic of third party CIs. 20:39:46 I would really like to see those for the drivers. 20:39:59 is this over-the-cloud deployment of each vendor in to a tenant, or under-the-cloud? Meaning: L3 adjacent, or L2 adjacent awareness (VXLAN, GRE, etc.) 20:40:15 nmangnezi I would also like to see one for the RedHat work. 20:41:30 m-greene Are you asking about the third party CI or the testing instance for the driver integration into the octavia API? 20:41:58 the latter… if I try to acquire a VE ‘lab’ or trial license.. 20:42:04 to drop into the multi-vendor lab 20:42:14 johnsom, third party CI? shouldn't centos/fedora based amphora run as a regular gate (voting or not) eventually? 20:42:27 m-greene: the more options provided, the more blogan can test before it hits the wild. 20:43:13 nmagnezi Yes, you are right, the RH tests would not need to be third party CIs. I mis-spoke there. 20:43:16 nmagnezi: it should, m-greene johnsom is talking about that it would be great if you can provide a private gate for testing octavia with you're driver 20:44:00 diltram: well, there's two things. provided a testbed for driver shim dev, and third-party CI. separate requests. 20:44:08 aside: the F5 plugin driver/agent still follows the legacy arch (pre-octavia), so not sure if that eliminates us from helping or driver testing 20:44:16 Ha, I am talking about both actually. Eventually getting a third party CI gate that tests the vendor driver. The other is helping blogan do the dev for the API 20:44:18 diltram, for load balancing we use haproxy, which is the default 20:44:20 dougwig: true 20:44:49 m-greene: drivers api will be compatible with nlbaas 20:45:14 still working towards hooking up a local F5 testbed to vote on commits (3rd party CI) 20:45:16 m-greene: the shim blogan is working on is to allow the neutron-lbaas v2 drivers to work in octavia, so it's very relevant. 20:45:16 nmagnezi: true but you're using different base image 20:45:26 unless you monkey around the lbaas plugin, and then you'll have some work to do. 20:45:36 dougwig: understood 20:45:45 (which we all do, so...) :) 20:45:53 diltram, that does not count as a 'driver', doesn't it? :) 20:46:51 nmagnezi: no, you can just use a specific gate like gate-neutron-lbaasv2-dsvm-scenario-{centos,fedora} 20:47:12 and configure it to build amphora image using {centos,fedora} 20:48:14 diltram, we will get there eventually, when i finish with https://review.openstack.org/#/c/331841/ , we will sure have that discussion - as it will be relevant and doable. 20:48:37 nmagnezi: edit the gate hook to build the fedora/redhat/centos amphora and use it instead of stock, then it's the standard gate job with a different name. 20:48:38 nmagnezi +! 20:48:55 dougwig +1 too 20:49:06 I can help with that once you are ready. 20:49:18 diltram, that the moment diskimage-create is capable of creating centos/fedora/ubuntu based amphoras but the agent should be adapted. 20:49:39 I finally was able to complete the policy so I will take a look into reviews 20:50:07 dougwig, right. same idea like we did with octavia/namespace gates for neutron-lbaas 20:50:11 so even if I will forgot about some PS just ping me 20:50:18 nmagnezi: *nods* 20:52:16 Ok, we have about eight minutes left. Any other topics today? 20:52:35 johnsom, btw if we already mention https://review.openstack.org/#/c/331841/ , i will try to resolve https://bugs.launchpad.net/octavia/+bug/1646125 in that patch but note that https://github.com/openstack/octavia/blob/master/elements/haproxy-octavia/os-refresh-config/configure.d/20-haproxy-selinux is not exactly an selinux policy 20:52:35 Launchpad bug 1646125 in octavia "Lack of SELinux policies prevents normal operation of a CentOS based amphora" [High,Confirmed] - Assigned to Nir Magnezi (nmagnezi) 20:52:59 Yes, that would be good to get cleaned up 20:53:05 i need to check if it's possible with semanage (i never tried it) 20:53:29 johnsom, selinux is active in the ubuntu based amphora? 20:53:55 nmagnezi No, I don't think ubuntu is running selinux 20:54:09 johnsom, wow :D 20:54:38 nmagnezi: ubuntu uses their own selinux thing if I'm right 20:55:06 Yeah, I was trying to remember what it was called. 20:55:30 Though we have not had to do any policy work for it on ubuntu. It is something we should strive for. 20:55:35 johnsom: the apparmor 20:55:43 Right, that was it 20:55:49 yay for consistency i guess :) 20:55:56 nmagnezi: +1 20:56:01 Yeah, tell me about it. 20:56:09 but true we should take a look into selinux/apparmor 20:56:37 Yep. I think nmagenzi has a good start for selinux on RH, so good stuff. 20:56:52 doing my best :) 20:56:58 johnsom: +1 20:57:21 Ok, just a few more minutes. Thanks again for all your work on the project! 20:58:13 #endmeeting