14:01:02 <enikanorov__> #startmeeting neutron lbaas
14:01:03 <openstack> Meeting started Thu Feb  6 14:01:02 2014 UTC and is due to finish in 60 minutes.  The chair is enikanorov__. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:01:04 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:01:04 <sbalukoff> Howdy! I am!
14:01:07 <openstack> The meeting name has been set to 'neutron_lbaas'
14:01:08 <edhall> hi
14:01:49 * pcm_ lurking
14:02:05 <enikanorov__> ok, lets start with the announcements
14:02:14 <vjay> hi
14:02:20 <avishayb> hello
14:02:23 <enikanorov__> and the most important is that gates are working more or less
14:02:41 <obondarev> \o/
14:02:48 <enikanorov__> so core team may spend more time on reviews
14:03:01 <enikanorov__> and also you actually can get +1 on your patches
14:03:06 <vjay> :-)
14:03:15 <evgenyf> Good!
14:03:15 <enikanorov__> and if it's -1 from Jenkins - you need to really dig into the logs
14:03:20 <enikanorov__> and not just recheck no bug :)
14:03:42 <baoli> Hi, rkukura
14:03:58 <enikanorov__> the second announcement is that feature proposal deadline is 18th of feb
14:04:16 <enikanorov__> if i'm not mistaken, so any new featuers implementation should be published before that
14:04:47 <vjay> what is published? merged to trunk or submitted to Gerrit?
14:05:01 <enikanorov__> pushed to gerrit. if it's on review alredy - that's fine
14:06:14 <enikanorov__> we have whole line of patches under development on all three major features
14:06:21 <enikanorov__> l7, lb instance, ssl
14:06:59 <enikanorov__> the features are important and make a big part of lbaas API and functionality
14:07:46 <enikanorov__> in my opinion there are very little chances that what we are working on right now will be merged in Icehouce
14:08:33 <enikanorov__> but at the same time we'd like to give users some advancement in lbaas service over what whas in Havana
14:09:11 <enikanorov__> so there is an idea to make a 'downstream' version of neutron (focused on lbaas plugin and drivers)
14:09:45 <vjay> what does that mean? can you elaborate?
14:09:49 <enikanorov__> yep
14:10:24 <enikanorov__> that supposed to be the merge of current master + exsiting features on review, which are stable and tested
14:10:51 <enikanorov__> so the branch could be packed and given to users for early adoption
14:11:14 <enikanorov__> so I'd like to ask if that makes sense to you guys?
14:11:28 <vjay> who will track and make this release happen?
14:11:29 <obondarev> who will be using that downstream version? Are therte such requests from someone?
14:11:48 <evgenyf> When should that happen?
14:12:13 <avishayb> I think some of Radware  customers are candidates
14:12:14 <enikanorov__> vjay: i guess i'm going to do that if there is such demand
14:12:22 <enikanorov__> now i'm exploring the need for that
14:12:51 <enikanorov__> of course this is some additional work that needs to be done
14:12:56 <avishayb> we can not come with nothing in lbaas for H version
14:13:03 <enikanorov__> however it may be beneficial for some vendors/customers
14:14:23 <enikanorov__> but having downstream version doesn't means every unmerget lbaas patch gets in
14:14:42 <enikanorov__> so we need some kind of policy for that (of course focused on quality)
14:14:55 <obondarev> are there any examples of such downstream versions in any other OS projects?
14:15:22 <enikanorov__> obondarev: in fact many adopters use whiole OS in such manner
14:16:00 <evgenyf> enikanorov__: are we planning to do this version before or after a freeze of Icehouse
14:16:00 <enikanorov__> I'm trying to understand will it be beneficial for development speed
14:16:05 <obondarev> yeah, I know, my question was about kind of official downstream version
14:16:15 <enikanorov__> as we'll need to work with two repos
14:16:44 <enikanorov__> evgenyf: we don't have deadlines, although what I'm talking about is not something decided
14:16:50 <enikanorov__> it's just an idea
14:16:59 <enikanorov__> i'm going to discuss it with Mark
14:17:28 <vjay> How will fixes get into this branch?
14:17:29 <enikanorov__> as we don't wan't to position it as 'here, have that, because neutron could not make anything in Icehouse'
14:18:16 <vjay> till when we will support this?
14:18:36 <enikanorov__> vjay: i think if a patch passes Jenkins and the author has tests to verify the patch, it will be a candidate for inclusion
14:18:44 <enikanorov__> by saying tests i mean integration tests
14:18:51 <enikanorov__> (scenarios)
14:19:27 <enikanorov__> vjay: good question. It needs to be decided
14:20:14 <enikanorov__> meanwhile it would be good if you could ask potential customers if they will be willing to try such kind of stuff.
14:21:13 <vjay> well, i know of two types of customers. one that dont want to move from havana to icehouse. Others who wants to jump from grizzly directly to icehouse.
14:21:40 <vjay> using this version will be very tricky
14:21:50 <obondarev> a question: what about horizon? do we need a downstream horizon version for new features merged to downstream neutron?
14:22:07 <enikanorov__> ok, basically that's the argument against the idea :)
14:22:16 <vjay> will this branch be kept in sync with neutron changes in icehouse?
14:22:24 <vjay> changes == fixes
14:22:25 <enikanorov__> obondarev: i don't think so, CLI would be enough
14:22:40 <enikanorov__> vjay: yes
14:22:58 <vjay> and it will be cut on top of havana?
14:23:01 <obondarev> enikanorov__: is CLI really ok for customers?
14:23:04 <vjay> sorry
14:23:06 <vjay> icehouse
14:23:44 <enikanorov__> vjay:  i think it will be following current master, because we are not targeting particular release
14:24:01 <enikanorov__> it matters how we distribute the code
14:24:20 <vjay> then it will be tricky.
14:24:24 <obondarev> enikanorov__: ok, so for CLI we need a downstream neutron client version - right?
14:24:24 <enikanorov__> it could be a mere diff applied to a certain revision of neutron's master
14:24:31 <enikanorov__> obondarev: yes
14:24:57 <enikanorov__> vjay: it's not very simple, but we can try to make it simple
14:25:18 <obondarev> enikanorov__: things are getting complicated :)
14:25:30 <enikanorov__> well, not really
14:25:37 <vjay> hmm...it is a long shot... as you said we need to first verify the need.
14:25:45 <enikanorov__> to test a new feature in the devstack you only need to provide a link to a repo
14:25:48 <enikanorov__> in your localrc
14:26:02 <enikanorov__> so using downstream version should not be more complex
14:26:20 <edhall> so what specific functionality will be added to this branch?
14:27:24 <enikanorov__> edhall: based on review results. it could be a common case when feature is not getting approved because of some minor concerns or feature freeze or ...
14:27:39 <enikanorov__> in this case if we are sure it's working - we add it to our branch
14:29:46 <enikanorov__> any more questions on the downstream idea?
14:30:30 <evgenyf> Let's here Mark's opinion
14:30:36 <sbalukoff> Expansion on edhall's question: Are there specific features being considered to add to this branch right now?
14:30:37 <evgenyf> hear*
14:31:15 <enikanorov__> sbalukoff: i think we don't have major feature that is ready for that
14:31:24 <sbalukoff> Ok.
14:31:31 <enikanorov__> loadbalancer instance may be, but it's not a big addition to functionality
14:32:08 <enikanorov__> also, we'll have 3 months between feature freeze and the time new features will be reviewed for merging
14:33:19 <vjay> enikanorov__: just to double confirm. there is no chance that ssl, l7 will make it to icehouse. is that right?
14:33:58 <enikanorov__> i didn't say no chance, but i don't think we will manage to polish those features
14:34:18 <edhall> so this is essentially a way to progress in the face of the Feb 18th feature freeze?
14:34:45 <sbalukoff> That's what it sounds like to me, too.
14:34:47 <obondarev> feature freeze is Feb 28 I think
14:35:22 <enikanorov__> yes, but our deadline is closer
14:35:31 <obondarev> so we have a bit more time to polish
14:35:54 <obondarev> what is our deadline?
14:35:58 <enikanorov__> right, our objective is to publish the code before 18th :)
14:36:04 <enikanorov__> not have it merged
14:36:35 <enikanorov__> we'll have ~2 additional weeks to address review comments
14:36:44 <enikanorov__> but it's still lot's of work
14:36:47 <obondarev> right
14:37:29 <obondarev> it's all depends on the review speed
14:37:34 <obondarev> from core reviewers
14:37:44 <enikanorov__> i think it also depends on development speed right now
14:37:56 <enikanorov__> as we don't have working implementations of l7 and ssl yet
14:38:03 <obondarev> right now yes
14:39:25 <obondarev> so let's move to the statuses on features?
14:39:26 <enikanorov__> let's briefly discuss the features under development
14:39:37 <enikanorov__> i'll start with lb instance
14:40:04 <enikanorov__> the patch is ready, obondarev is working on applying instance approach to haproxy
14:40:17 <obondarev> I've uploaded a patch wich adopts lb instance for agent scheduling
14:40:19 <enikanorov__> so users will be able to create multiple pools in haproxy
14:40:42 <enikanorov__> these are prerequisities for L7 rules
14:40:54 <obondarev> second patch will adopt lb instance for lbaas agent itself, I'm working on it
14:42:02 <evgenyf> my next steps for SSL are: writing db layer unitests and fixing all review comments
14:42:12 <enikanorov__> avishayb: obondarev: what's the status of l7? do we have CLI for l7 resources?
14:42:31 <avishayb> I have updated the wiki - https://wiki.openstack.org/wiki/Neutron/LBaaS/l7
14:42:55 <evgenyf> BTW, I have undated the WIKI page too, https://wiki.openstack.org/wiki/Neutron/LBaaS/SSL
14:42:59 <enikanorov__> evgenyf: good
14:43:07 <avishayb> its in sync with the code now. Begining next week I will start to react on reviews
14:43:15 <avishayb> No CLI yet
14:43:20 <enikanorov__> avishayb: i see
14:43:29 <enikanorov__> evgenyf: i have another concern on ssl API
14:43:33 <obondarev> avishayb: I put some new comments on the patch. Also will be great if you can address/answer comments on the previous patch sets
14:43:55 <enikanorov__> it seems complex to me, so I'd like to see some ability to use simplified workflow
14:44:00 <avishayb> obodarev: I will
14:44:01 <obondarev> it's really hard to review if you don't know what you agrre and what not and why
14:44:02 <enikanorov__> i'll explain what i mean
14:44:03 <evgenyf> There is beagaviour description for vip/pool/ssl policy protocole mismatches. Guys, please have a look..
14:44:13 <evgenyf> *behaviour
14:45:04 <enikanorov__> by saying 'complex API' i don't mean it needs to be changed, but instead it should provide ability to do things on less steps
14:45:15 <obondarev> avishayb: thanks
14:45:24 <enikanorov__> say, create a vip with 1 command, providing all ssl attributes
14:45:34 <enikanorov__> evgenyf: do you think it's possible?
14:45:54 <enikanorov__> 1 command = 1 rest call
14:47:06 <evgenyf> enikanorov__: Do you mean using CLI?
14:47:31 <enikanorov__> no, i'd prefer to see it in the API
14:48:11 <enikanorov__> it's like loadbalancer instance, you can create one, and then add pool to it, or you can create a pool and lb instance will be created for the pool automatically
14:48:34 <enikanorov__> can we apply the same approach to policies/certs?
14:49:21 <evgenyf> you mean applying default SSL poliy and certs. to vip on creation stage?
14:50:09 <vjay> default cert?
14:50:13 <enikanorov__> no
14:50:24 <enikanorov__> i mean i could provide cert on vip create
14:51:04 <evgenyf> I do not see a reason for doing that.. can you elaborate ?
14:51:05 <enikanorov__> i'm just thinking of the terrible amount of parameters i need right now to setup a ssl vip
14:51:44 <enikanorov__> ok, i'll review the API again and will explain more precisely
14:52:08 <evgenyf> enikanorov__: Ok, let's discuss it
14:52:32 <enikanorov__> ok, that's all i wanted to discuss today
14:52:43 <enikanorov__> doesn anyone have questions/items to discuss?
14:53:12 <evgenyf> Please review SSL wiki page for new details
14:53:22 <obondarev> lbaas scenario test is in good shape, should be merged soon
14:53:41 <vjay> is anyone working on sslconnect+haproxy?
14:54:12 <enikanorov__> vjay: stunnel?
14:54:26 <vjay> sorry, yes stunnel
14:55:25 <enikanorov__> it's me, but i have not made much progress on that particular item as I'm working with evgenyf on ssl API patch
14:56:24 <vjay> ok.
14:57:55 <enikanorov__> ok, if theres' no other questions then let's wrap up
14:58:02 <enikanorov__> thanks for joining everyone
14:58:35 <vjay> bye.
14:58:42 <sbalukoff> Seeya!
14:58:43 <obondarev> bye all
14:58:48 <s3wong> bye
14:58:51 <evgenyf> bye
14:58:52 <enikanorov__> see you all
14:58:57 <enikanorov__> #endmeeting