21:00:07 <danwent> #startmeeting quantum
21:00:07 <openstack> Meeting started Mon Dec 17 21:00:07 2012 UTC.  The chair is danwent. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:00:08 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:00:10 <openstack> The meeting name has been set to 'quantum'
21:00:17 <danwent> so…. good news/bad news...
21:00:29 <danwent> good news is that the meeting agenda was really easy to update this time around...
21:00:40 <enikanorov> hi
21:00:44 <danwent> bad news is that this is because its largely unchanged from last time
21:00:57 <danwent> meaning we didn't get all that much accomplished last week.
21:01:34 <danwent> its tough around the holidays, but remember that the G-2 deadline is Jan 22nd, which doesn't leave a lot of time once people get back from the holidays, so we really need to be clearing out our feature review backlog now.
21:02:06 <danwent> ok, enough soap-box preaching, let's get down to business
21:02:17 <danwent> #topic announcements
21:02:35 <danwent> #info thanks to mnewby, we now have a base layer of quantum tests integrated into tempest
21:02:52 <danwent> #info there is a devstack patch to simplify running these test, see: http://wiki.openstack.org/QuantumDevstack
21:02:58 <danwent> any other announcements?
21:03:19 <danwent> #topic quantum documentation
21:03:20 <Nachi> quantum-gate is still not working (Bad news)
21:03:32 <danwent> Nachi: we'll get to gating soon :)
21:03:46 <Nachi> danwent: I hope so
21:04:04 <danwent> on the docs, we are largely where we were last week.  Just want to draw attention to two reviews
21:04:05 <danwent> https://bugs.launchpad.net/openstack-manuals/+bugs?field.tag=quantum
21:04:10 <gongysh> It seems we are inside, so there is no need for us to build gate.
21:04:15 <danwent> in particular
21:04:17 <danwent> https://review.openstack.org/#/c/17570/
21:04:28 <danwent> https://review.openstack.org/#/c/13759/
21:04:47 <danwent> gongysh: i think this patch needs some discussion from you before more review: https://review.openstack.org/#/c/17774/
21:05:14 <danwent> gongysh: i think we need the gate to keep out barbarian hordes of developers who will destroy quantum :)
21:05:34 <danwent> Anything else on docs?
21:05:46 <annegentle_> for 17570, are you all ready for review again gongysh ?
21:05:48 <danwent> I have noticed people being really helpful about responding to questions on Answers on launchpad
21:06:28 <danwent> this is great.  and when a question is answered, its always good to think about whether it SHOULD have been answered by the docs, and if it was not in the docs, to create a doc bug for it.
21:06:58 <gongysh> annegentle: yes
21:06:59 <danwent> our goal should always be to make ourselves irrelevant in terms of needing to answer user questions :)
21:07:09 <salv-orlando> annegentle_: 17570 fell of my radar, will review later today
21:07:18 <danwent> in particular, i've noticed gongysh doing a great job responding to questions on LP Answers, so thanks Yong!
21:07:25 <annegentle_> salv-orlando: ok, great.
21:07:49 <gongysh> danwent: that is my first job of every day.
21:08:00 <danwent> gongysh: i can tell, very helpful for the community.
21:08:05 <annegentle_> salv-orlando: and I'll look at 13759
21:08:07 <danwent> ok, let's move past docs then.
21:08:12 <annegentle_> wait,
21:08:20 <danwent> #topic grizzly-2 status
21:08:21 <annegentle_> I wanted to ask about the Quantum WADL generator?
21:08:38 <danwent> annegentle_: sure, go ahead.  gongysh is someone from your team working on that?
21:08:52 <gongysh> annegentle: zyluo is working on it.
21:08:52 <annegentle_> so this page got on my radar http://wiki.openstack.org/Quantum/API/WADL
21:09:02 <zyluo_> I recently wrote a spec
21:09:02 <annegentle_> and other teams are interested if you get it working - how far along are you?
21:09:22 <annegentle_> zyluo_: it would be super helpful to a bunch of projects, seriously neat.
21:09:29 <annegentle_> #link http://wiki.openstack.org/Quantum/API/WADL
21:09:39 <zyluo_> well I'm currently looking on what Nova has done with automated tests.
21:10:02 <zyluo_> Then I'll fill in some of gongysh 's questions and start working on that bp
21:10:15 <annegentle_> zyluo_: fabulous, thanks.
21:10:25 <zyluo_> Targeted to G3
21:10:29 <danwent> cool stuff!
21:10:39 <annegentle_> Ok, carry on, thanks for letting me poke my head in.
21:10:52 <danwent> annegentle_, as always, good seeing you!
21:10:56 <annegentle_> :)
21:11:14 <danwent> #info https://launchpad.net/quantum/+milestone/grizzly-2
21:11:42 <danwent> looks pretty similar to last week, not a lot of progress in the past week.
21:12:08 <danwent> in this meeting, i'd really like to identify 3 features we think we can get merged this week, and get a set of people committed to doing so.
21:12:14 <danwent> https://blueprints.launchpad.net/quantum/+spec/quantum-gate
21:12:27 <danwent> sounds like we're still seeing differences between our setups, and the gate on this one, huh?
21:12:33 <Nachi> danwent: yes
21:12:43 <danwent> ETA on fix is unknown, so probably uncertain if it will merge this week?
21:12:55 <Nachi> danwent: yes X(
21:13:14 <garyk> Nachi: is there anyway which we can help here?
21:13:21 <danwent> ok.  Nachi, so you're working on this and the security groups stuff at the same time.  wondering if we can offload
21:13:24 <danwent> exactly :)
21:13:58 <danwent> from amotoki's note, it sounds like there is additional unit test dev requirement for the sec group stuff
21:13:58 <nati_ueno> sorry disconnected
21:14:11 <nati_ueno> ah topic is in sec group?
21:14:17 <garyk> nati_ueno: is there anyway which we can help?
21:14:23 <danwent> nati_ueno: what was the last thing you saw?
21:14:24 <nati_ueno> I agree with amotoki. I'll work on unit tests in this week
21:14:33 <nati_ueno> -> nati_ueno: is there anyway which we can help?
21:14:34 <danwent> garyk: Nachi: is there anyway which we can help here?
21:14:34 <danwent> [1:13pm] danwent: ok.  Nachi, so you're working on this and the security groups stuff at the same time.  wondering if we can offload
21:14:36 <danwent> [1:13pm] danwent: exactly
21:14:37 <danwent> [1:13pm] Nachi left the chat room. (Read error: Connection reset by peer)
21:14:38 <danwent> [1:13pm] nati_ueno joined the chat room.
21:14:39 <danwent> [1:13pm] danwent: from amotoki's note, it sounds like there is additional unit test dev requirement for the sec group stuff
21:14:41 <nati_ueno> danwent: Thanks
21:15:32 <danwent> nati_ueno: would someone else be able to pitch in and help try to understand the gating failures?
21:15:39 <nati_ueno> garyk: I'm working on create local CI with devstack-gate script. But I got kernel panic from VM. May be my hardware problem or config problem.
21:15:54 <mestery> nati_ueno: Let me know if I can help here too.
21:16:06 <zyluo_> nati_ueno, me2
21:16:09 <mestery> nati_ueno: I have some cycles and machines for this too.
21:16:18 <nati_ueno> Thanks folks. Could you also trying to create local CI too?
21:16:29 <danwent> i'd love it if more people got involved in the CI-side of things for quantum.
21:16:32 <mestery> nati_ueno: will do
21:16:38 <garyk> nati_ueno: can you please post the instrauctions and then all intersted can help check it out
21:16:41 <danwent> nati_ueno: can you send out some pointers to the list on this?
21:16:46 <danwent> :)
21:16:48 <mestery> :)
21:16:57 <nati_ueno> sure!
21:17:12 <danwent> ok, next up, markmcclain
21:17:13 <danwent> https://blueprints.launchpad.net/quantum/+spec/quantum-db-upgrades
21:18:03 <danwent> markmcclain: ping?
21:18:12 <markmcclain> hope to push tonight/tomorrow morning
21:18:29 <nati_ueno> mestery: garyk: here is instructions of local CI gating https://github.com/openstack-infra/devstack-gat
21:18:39 <nati_ueno> zyluo_: also
21:18:53 <nati_ueno> sorry typo https://github.com/openstack-infra/devstack-gate
21:18:54 <garyk> danwent: markmcclain i saw that there is a patch in common for db support. is this something to take into account for the upgrades?
21:19:13 <danwent> markmcclain: how complex is the patch?  is it feasible to review this week?
21:20:23 <markmcclain> not terribly complex.. should be reviewable this week
21:20:35 <markmcclain> more just making sure it works for the various plugins
21:21:09 <danwent> markmcclain: ok.  i forget, who volunteered to be the core devs here?  I think i did.
21:21:24 <markmcclain> I thought it was you and salv-orlando
21:21:41 <danwent> i'm stuck in a mandatory company training all day tomorrow + wed (urgh), but can review on either side of the main workday
21:21:43 <salv-orlando> I was one of the two
21:22:18 <danwent> ok, well, if another core dev beyond salv-orlando comes in and reviews it more quickly, don't feel you need to wait for me.
21:22:38 <danwent> ok, i'm putting this on my "hit list" of items to merge this week.  feeling lucky :)
21:22:49 <danwent> https://blueprints.launchpad.net/quantum/+spec/quantum-security-groups-iptables
21:23:13 <danwent> nati_ueno: from email discussion, sounds like we should WIP this until we get some more unit test coverage on the agent/iptables-manager side?
21:23:22 <danwent> was I understanding amotoki correctly?
21:23:24 <nati_ueno> danwent: Yes I agree
21:23:38 <danwent> ok, please put review in WIP.
21:23:45 <danwent> https://blueprints.launchpad.net/quantum/+spec/lbaas-restapi-tenant
21:23:53 <danwent> looks like we have a +2 from salv-orlando on this one.
21:24:00 <danwent> can we FINALLY merge this?
21:24:01 <enikanorov> danwent: i think the patch (both patches, in fact) is ready and is looking good. waiting for markmcclain to approve.
21:24:03 <nati_ueno> sure
21:24:17 <enikanorov> as he was leaving some coments for the latest version
21:24:18 <salv-orlando> yes, but this patch depends on another one where markmcclain had a comment to address
21:24:26 <danwent> salv-orlando: link?
21:24:41 <danwent> sub-resoruces?
21:24:43 <enikanorov> https://review.openstack.org/#/c/16888/
21:24:43 <markmcclain> https://review.openstack.org/#/c/16888/5
21:24:56 <danwent> yup
21:25:02 <salv-orlando> I'm not the fast clicker I used to be :)
21:25:14 <enikanorov> I think Oleg has responded to the comments
21:26:12 <danwent> markmcclain: ok, please try to respond to this today, so we can get the two patches rolled in.
21:26:34 <danwent> enikanorov: do you know who the second core dev is going to be on the main patch?  I know its had a lot of reviewers, but who else is actively engaged at this point?
21:26:53 <markmcclain> missed the responses earlier.. I'll respond
21:26:54 <garyk> danwent: i was looking at them. i can look again
21:27:17 <markmcclain> I'll be on the 2nd patch too, just wanted first to last
21:27:18 <danwent> garyk, markmcclain: great, thanks
21:27:19 <markmcclain> *land
21:27:38 <danwent> ok, so I am putting this one on my hit-list too :)
21:27:55 <danwent> which means i'll be pestering you if you're a reviewer or a dev and things seem stuck :)
21:28:05 <danwent> garyk https://blueprints.launchpad.net/quantum/+spec/vif-plugging-improvements
21:28:23 <danwent> i think i probably owe you feedback on this one
21:28:30 <danwent> did you end up merging the nova side reviews?
21:28:31 <garyk> danwent: final patch is in review. got -1. there are/were expectation that i rewrite nova's networking.
21:28:58 <garyk> i need you to take a look and get your 2 cents. i think it is a reasonable approach the one i have taken
21:29:14 <danwent> ok, will do that later today
21:29:21 <garyk> danwent: got 2 out of the 3 merged. thanks!
21:29:50 <danwent> https://blueprints.launchpad.net/quantum/+spec/lbaas-plugin-api-crud
21:30:06 <danwent> looks like leon has been making progress here
21:30:25 <danwent> still going through feedback & revision though.
21:30:34 <enikanorov> yes. but still the patch has to be presented properly :)
21:31:16 <danwent> enikanorov: can you elaborate?
21:32:01 <enikanorov> I think salv-orlando has made a comment regarding how Leon should have used Oleg's patches
21:32:27 <danwent> ok, just wanted to make sure your comments were already captured in the existing review comments.
21:32:55 <enikanorov> in fact we're already kind make some integration testing with this code
21:33:04 <enikanorov> *kind of
21:33:06 <danwent> my spidey-sense says that this review is probably more than a week out, but hopefully we can still make good progress on reviews.
21:33:23 <danwent> enikanorov: great.  sounds like you're giving helpful feedback to leon
21:33:28 <danwent> https://blueprints.launchpad.net/quantum/+spec/quantum-service-type
21:33:40 <danwent> salv-orlando: things seem to be progressing well here?
21:34:06 <salv-orlando> yes
21:34:16 <salv-orlando> I need to address the comments
21:34:17 <danwent> is it fair to shoot for a merge this week?
21:34:42 <salv-orlando> I think it's fair; the only concern might lay around how adv services should use the service type
21:34:53 <salv-orlando> but I think I can work that out with the reviewers
21:35:08 <salv-orlando> I also need to push a CLI patch for that API extension, but that should be easy
21:35:15 <danwent> salv-orlando: ok.  i agree that this abstraction may need to change once we gain more experience with it by actually writing services
21:35:27 <danwent> ok, so my official hit-list for this week is:
21:35:32 <danwent> https://blueprints.launchpad.net/quantum/+spec/quantum-db-upgrades
21:35:38 <danwent> https://blueprints.launchpad.net/quantum/+spec/quantum-service-type
21:35:45 <danwent> https://blueprints.launchpad.net/quantum/+spec/lbaas-restapi-tenant
21:35:57 <danwent> (where the last one includes both the sub-resources patch and the main patch)
21:36:01 <danwent> did I miss anything?
21:36:12 <garyk> vif plugin?
21:36:36 <garyk> danwent: it is more in the scope of nova
21:37:02 <danwent> garyk: sure.  ok, please send me the link for that.  yeah, its in nova, so not sure how much control we have over merging timelines :)
21:37:09 <danwent> i also wanted to talk about https://blueprints.launchpad.net/quantum/+spec/lbaas-restapi-tenant
21:37:24 <garyk> dansmith: https://review.openstack.org/#/c/17747/
21:37:58 <danwent> as its important that we get at least one open source load balancing plugin in during grizzly, and I don't want it to land late in G-3, as we would lose the ability to get experience using it, and potentially make important adjustments.
21:38:36 <danwent> i know there have been a lot of threads going on about the design of an agent-based plugin for LBaaS
21:38:59 <danwent> i'm not sure if people are coalescing around a single design, or planning on creating different plugins
21:39:14 <danwent> so I wanted to do a status check and make sure we were making forward progress.
21:39:30 <enikanorov> in fact, agent-based design is being prototyped already
21:39:58 <ilyashakhat> and we will have one driver for HAProxy
21:40:02 <danwent> enikanorov: i figured, which is great.  do you know if youcef and others are planning on using this same code, or are they plannign on doing somethign else?
21:40:02 <enikanorov> we've got some working code
21:40:23 <danwent> enikanorov: is this code publicly available someplace (even if not ready for a review?)
21:40:29 <enikanorov> well, i hope when we have time for this, it all will be discussed
21:40:34 <enikanorov> and agreed upon
21:40:38 <Youcef> danwent: yes, we decided early on to have one lbaas plugin for everybody with multiple drivers
21:41:01 <enikanorov> currently we discuss it on ML
21:41:14 <enikanorov> since other places like quantum meetings are busy with other action items
21:41:23 <danwent> Youcef: yes, trying to understand whether we are close to agreement on the "one" agent-based design.
21:41:48 <danwent> so, sounds like there are still active issues that need to be resolved via discussion on ML?
21:42:18 <enikanorov> i think there are some issues
21:42:22 <ilyashakhat> there's an open question regarding scheduling loadbalancer objects on loadbalancing devices
21:42:32 <enikanorov> it that we haven't payed too much attention to this lately.
21:42:35 <Youcef> danwent: yes, the agent/driver interface with the lbaas plugin is yet to be discussed and designed.
21:43:03 <enikanorov> Youcef: we have wiki page on that with a proposal of such interface
21:43:45 <danwent> Youcef, enikanorov: ok.  i agree we don't need to try and design it in this meeting.  I'd just like to have a weekly check-point where we can check-in on progress and high-light the key questions we're facing here, as I know a lot of peopel are anxious to see progress here.
21:43:54 <Youcef> enikanorov: was it updated after scheduling discussion and service_types clarifications?
21:44:07 <ilyashakhat> Youcef: not yet
21:44:29 <danwent> ok, moving on.
21:44:46 <enikanorov> i think schedulingand service types are a bit aside from driver interface
21:44:48 <danwent> amotoki, nati_ueno : looks like horizon BPs have been targeted for G-2 and given a priority, thanks for that.
21:44:49 <enikanorov> ok
21:45:12 <danwent> enikanorov + Youcef: i'll look to both of you in future meetings for an update on the status of that BP.  thanks.
21:45:20 <amotoki> danwent: gabriel just approved them :-)
21:45:38 <danwent> nati_ueno: given your load on sec-groups and gating, is it realistic to target two more BPs for horizon in G-2?
21:45:45 <danwent> sounds like you'll be working through christmas
21:45:53 <nati_ueno> danwent: two BP?
21:46:03 <danwent> nati_ueno: ah, is it only the L3 one that is G-2?
21:46:14 <nati_ueno> danwent: Yes. I think it is doable
21:46:31 <danwent> ah, topo stuff is G-3, i see.  ok, seems reasonable :)
21:46:46 <danwent> Ok, anything else for G-2?
21:46:59 <garyk> danwent: nati_ueno: is there a design for this. i am not sure that the l3 agent is that simple. the networking here may be challenging
21:47:23 <nati_ueno> garyk: horizon router pane?
21:47:24 <danwent> garyk: ?
21:48:12 <garyk> danwent: nati_ueno: the l3 agent. maybe i am coinfused. past my be time
21:48:27 <garyk> i was thinking multinode
21:48:37 <danwent> garyk: nati_ueno is talking about updating horizon web-gui to be able to talk to L3 quantum APIs
21:48:50 <garyk> danwent: my bad (past my bedtime)
21:48:55 <nati_ueno> garyk: AH but it should be discussed too
21:48:59 <danwent> basically, do same ops at router-* CLI commands.
21:49:11 <garyk> undertsood
21:49:33 <danwent> nati_ueno: is there another BP your'e talking about?
21:49:45 <danwent> i dont' see it on grizzly-2 page
21:50:12 <gongysh_> quantum scheduler?
21:50:15 <nati_ueno> https://blueprints.launchpad.net/quantum/+spec/quantum-scheduler
21:50:23 <nati_ueno> Ah but is is not targeted to G2 yet
21:50:40 <gongysh_> I have two patches for it.
21:50:45 <danwent> yes, i'm a bit worried about pouring more items into G-2 until we are in better shape of merging what we have.
21:50:48 <gongysh_> I hope it can go into G2.
21:50:59 <enikanorov> wow. interesting
21:51:14 <enikanorov> that's what we planned to introduce for adv services as well
21:51:16 <cdub_> is there any reuse possibllty forlbaas scheduler?
21:51:20 <danwent> gongysh_: ok, let's look at patches, but based on previous discussion, it seemed quite involved.
21:51:20 <nati_ueno> so we need to agree the spec
21:51:22 <enikanorov> sure
21:51:41 <nati_ueno> forlbaas scheduler ?
21:51:46 <gongysh_> yes, codes can help spec.
21:51:51 <danwent> https://docs.google.com/document/d/1TJlW0_tMpeENA_ia38fvRu7ioKRt9fsWXBjivwd1mMw/edit
21:52:06 <danwent> i don't ever remember seeing replies to a lot of my questions on the spec
21:52:48 <garyk> i have issues with the scheduler and the l3 agent. not 100% sure it will work with the routing.
21:52:57 <cdub_> nati_ueno: for lbaas scheduler ... "there's an open question regarding scheduling loadbalancer..."
21:53:16 <nati_ueno> cdub_: Thanks
21:53:28 <enikanorov> still, scheduling is a general mechanism for balancing load. sooner or later lbaas will have to use it
21:53:32 <danwent> ok, so let's try to get to clarity on the existing spec.  if we are able to do that soon, getting this in G-2 may be possible.
21:54:02 <danwent> one issue is that quantum folks use the term "scheduling" to mean several different things in my experience :)
21:54:02 <gongysh_> Danwent: I will answer your questions soon.
21:54:23 <danwent> gongysh_: ok, thanks.  let me take another look at the spec as well, as I remember being pretty uncertain about a few things.
21:54:37 <danwent> ok, are we able to wrap up the G-2 section of the meeting?
21:54:38 <nati_ueno> so we should hire same definition with nova-scheduler
21:54:40 <gongysh_> danwent: sometimes, codes are better.
21:55:00 <gongysh_> danwent: u can see my patches. both server and client.
21:55:03 <enikanorov> nati_ueno: agree.
21:55:04 <danwent> gongysh_: i agree that having code in addition to a spec is nice.  but we need a spec to be clear on goals and high-level design.
21:55:30 <danwent> #topic open discussion
21:55:43 <danwent> just want to call people's attention to the list of "Open Items Ready to be Claimed"
21:55:44 <garyk> danwent: is there a meeting next week?
21:55:47 <danwent> on the agenda
21:55:51 <danwent> garyk: that is item #2 :)
21:56:05 <garyk> ok
21:56:22 <danwent> if you're looking to find a new project to work on for quantum, please take a look at this list
21:56:34 <danwent> second item is that next monday is christmas eve
21:57:01 <danwent> which means attending a meeting will be tough for a lot of us.  i also expect the main openstack team meeting on christmas to be canceled.
21:57:16 <danwent> i'd suggest we replace a meeting with check-in emails on any important topics we think can't wait until the next week.
21:57:29 <mlavalle> danwent: any chance of getting some feedback from you regarding additional tempest tests development?
21:57:41 <danwent> but if others want to meet on IRC, that's fine.  just realize that many folks will not be able to attend.
21:58:00 <danwent> mlavalle: its in my long queue of emails to respond to, sorry
21:58:10 <mlavalle> np
21:58:34 <mlavalle> just want to make sure doesn't fall through the cracks
21:58:42 <danwent> mlavalle: i personally would love to see some of the key use cases show here (http://docs.openstack.org/trunk/openstack-network/admin/content/use_cases.html) coded up as tempest use cases and included in the automated tests that run on each build.
21:59:07 <mlavalle> danwent: ok, will take a look
21:59:29 <danwent> the unit tests do a fairly good job today in flexing the API code, so I think the most value first thing in terms of tempest tests would be replicating common deployment models.
21:59:46 <danwent> the real value of tempest in my mind is that it operates across nova, quanutm, keystone, etc.
21:59:50 <mlavalle> understood
21:59:52 <danwent> ok, any other open discussion?
22:00:14 <danwent> otherwise, we'll cancel the IRC meeting for next week.
22:00:50 <garyk> merry xmas and a happy new year!!!
22:00:57 <danwent> ok, thanks folks.  happy holidays, hope we wrap up some reviews this week, get a relaxing break and come engergized to finish up g-2 :)
22:01:00 <enikanorov> thanks :)
22:01:04 <salv-orlando> bye!!
22:01:05 <danwent> #endmeeting