21:02:12 <danwent> #startmeeting
21:02:13 <openstack> Meeting started Mon Jul  9 21:02:12 2012 UTC.  The chair is danwent. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:02:14 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:02:25 <danwent> gongys: hi
21:02:31 <rkukura> hi
21:02:53 <danwent> #link agenda: http://wiki.openstack.org/Network/Meetings
21:03:15 <danwent> #topic Folsom-2 recap
21:03:19 <garyk> hi
21:03:33 <danwent> overall, a very strong effort from the community.  we had a lot of people really step up, which was great.
21:03:42 <danwent> have a few things to wrap up
21:04:05 <ncode> o/
21:04:06 <danwent> dhcp is about ready to go in, though there are some issues that we're deferring to separate bugs
21:04:08 <danwent> https://review.openstack.org/#/c/9064/9
21:04:33 <danwent> gongys: I was mainly looking to get your thoughts on whether there was anything we couldn't defer to seperate bugs.
21:04:46 <danwent> gongys: where you able to test with null firewall driver in your setup?
21:05:14 <gongys> not. I am testing with firewall driver.
21:05:16 <danwent> gongys: i filed this to track the issue: https://blueprints.launchpad.net/quantum/+spec/expose-dhcp-server-ip
21:05:34 <gongys> That's fine.
21:05:48 <gongys> I think it can be dealt with by multi-host dhcp.
21:06:05 <gongys> or other multi-host features.
21:06:18 <garyk> did anyone open a bug regarding the dhcp tap device name?
21:06:44 <gongys> in which cases, we must expose dhcp server and gateway ips to other system, such as nova.
21:06:45 <danwent> garyk: not yet, I don't think.  please do.  there are a couple outstanding issues that we should file before actually approving
21:06:59 <gongys> Sorry, I have approved.
21:07:20 <danwent> gongys: yes.  gateway ip is already exposed for basic model, though multi-host may be tricky (would ahve to think more)
21:07:37 <danwent> gongys: no problem.  i'll just go back and look at the comments and make sure there's a bug foreach of them
21:07:38 <garyk> gongys: no problem
21:07:51 <markmcclain> yeah I've got four bugs to file
21:08:30 <danwent> markmcclain: ok, great.  please not that a few may already be filed (did it this morning).  just search for "dhcp" on that main folsom-3 page
21:08:44 <markmcclain> danwent: will do
21:08:44 <danwent> we also have the two devstack patches
21:09:00 <danwent> garyk's dhcp patch is just waiting on a devstack reviewer: https://review.openstack.org/#/c/9064/9
21:09:17 <garyk> danwent: ok, thanks
21:09:34 <danwent> (sorry, wrong patch)
21:09:35 <danwent> https://review.openstack.org/#/c/9362/
21:10:00 <danwent> plus we have arosen's patch for v2 support: https://review.openstack.org/#/c/9161/
21:10:14 <danwent> looks like he just revved it, so hopefully that's good to go today as well.
21:10:37 <danwent> So i think we can have all the basic stuff in today, assuming we can get dean to help us with a devstack core review.
21:11:08 <danwent> any other blockers for a basic setup?
21:11:26 <danwent> I want to make sure everyone is able to get a basic dev setup that they can use for developing F-3 items.
21:11:44 <zhanghua> that's cool!
21:11:52 <danwent> #topic Folsom-3
21:11:55 <danwent> https://launchpad.net/quantum/+milestone/folsom-3
21:12:28 <danwent> Due date is Aug. 16th.  4 days before i'm expecting my baby to be born.  Great timing :)
21:12:37 <garyk> sweet
21:12:40 <danwent> hehe, yeah
21:12:43 <gongys> nice.
21:12:45 <ncode> :D
21:12:51 <zhhuabj> :)
21:13:02 <gongys> We can defter the f3 4 days.
21:13:11 <danwent> we're obviously over subscribed on the release, but I wanted to put a lot there so people had a chance to see everything that's on our plate.
21:13:53 <danwent> as always, any major item needs to be proposed for a review a week before the branch date.
21:14:03 <danwent> which itself is two days before the release date.
21:14:20 <danwent> so all features should be in for review by Aug 7th.
21:14:47 <danwent> This week, take a look at the list, identify anything you think is missing, and put some thought into what you can work on during F-3
21:15:02 <danwent> Some areas where I have a lot of concern are around Horizon
21:15:33 <danwent> we made slow progress on the basic horizon integration during F-1 and F-2, and now arvind is out-of-office.
21:15:50 <danwent> we have a set of additional things to do in horizon beyond the base work as well (e.g., floating ips, etc.)
21:15:57 <danwent> so we could really use someone to step up there.
21:16:27 <danwent> Multi-host is another area that will need attention
21:16:54 <gongys> let me deal with mult-host.
21:16:58 <danwent> we split the multi-host dhcp work into another BP.  and we'll likely do the same for the mulit-host L3 stuff.  I'd really like to find someone planning on using multi-host who will put the time into it.
21:17:01 <danwent> gongys: ok, great.
21:18:03 <danwent> Another area were we need some though is a "notifications API", so that elements can register to learn about events in quantum, like when a port is created/updated/deleted or when a floating ip is allocated.
21:18:16 <danwent> this is related to some of the RPC work people have been thinking about.
21:18:35 <danwent> as much as possible, i'd like this to be a real notifications API, not just a bunch of adhoc RPC calls.
21:18:44 <garyk> danwent: i am working on the rpc with the scalable agents. hopefully will have linux brdige in soon
21:18:53 <danwent> this will let external systems (e.g., ceiliometer) integrate.
21:19:29 <danwent> garyk: cool, yes, I haven't looked at that code yet, so I'm not quite sure how it might relate to a notifications API, but it seems like at a high level you need to get notifications that new ports on being created on a particular host, is that correct?
21:19:40 <danwent> (i.e., the main goal is to avoid polling?)
21:19:50 <garyk> danwent: yes, that is correct.
21:20:16 <garyk> danwent: code was working with v1 now upgrading to v2 - ran into rpc issues
21:20:29 <danwent> ok… i'll try and take a look at this soon to figure out how it relates to our goals around notifications.
21:20:47 <danwent> garyk: ok, an this is using openstack common rpc, correct?
21:21:00 <garyk> danwent: yes, that is correct.
21:21:22 <danwent> ok, great
21:21:36 <garyk> based on https://review.openstack.org/#/c/9487/ and https://review.openstack.org/#/c/9166/
21:22:00 <danwent> And the final big area that doesn't have anyone assigned is API quotas and rate-limiting
21:22:39 <danwent> i'm sure there will be addtional areas where help is needed as well, but I'm just trying to call out major areas where no one has currently been working.
21:23:05 <danwent> so expect emails from me in the next week as I try to find people for certain blueprints + bugs.
21:23:40 <danwent> unlike Essex, folsom-3 is the last iteration in Folsom
21:24:01 <danwent> so anything that must be in Folsom needs to be completed in this iteration before we go into feature-freeze
21:24:26 <garyk> danwent: can we also dedicate some time to stable essex at the end of the meeting
21:24:31 <danwent> (and unlike essex, there's not a really long period for "feature freeze exceptions")
21:24:35 <danwent> garyk: good idea.
21:24:42 <danwent> will do it right after this section.
21:24:47 <garyk> tx
21:24:49 <danwent> any other questions/comments on Folsom-3
21:25:02 <gongys> Dan, Do you have any design in your mind about horizon integraton?
21:25:03 <rkukura> need some reviews on provider-networks
21:25:31 <danwent> gongys: the base work is fairly well flushed out and arvind has a branch (will send out link)
21:25:54 <danwent> but we still need to create BPs around additional functionality (e.g., floating IPs)
21:26:10 <danwent> #todo #danwent make sure additional horizon+quantum BPs are created.
21:26:30 <gongys> I thin we need get a spec out about what we will expect about horizon integration.
21:26:34 <danwent> rkukura: salv-orlando is not here, but I've asked him to focus on that review
21:26:42 <gongys> thin -> think.
21:27:24 <rkukura> danwent: thanks. Others welcome too
21:27:33 <danwent> gongys: yes.  I spoke with arvind about the base integraiton on the phone, and I believe that part is spec-ed out.  But the next steps beyond that are not well-speced.
21:27:43 <ncode> rkukura: I will look it today
21:28:10 <danwent> #topic Essex-stable
21:28:18 <danwent> garyk: all yours
21:28:47 <garyk> pedning review - https://review.openstack.org/#/c/8650/
21:29:32 <garyk> once this is done we need to add - a cherry pick for https://github.com/openstack/quantum/commit/438eda895c7e24113f116e503f36930c176ebe4d
21:30:13 <danwent> yeah, sorry, i've been ignoring those reviews to focus on F-2
21:30:17 <garyk> can anyone think of any other fixes that need to go in. the sooner we can get tarballs ready the better.
21:30:50 <garyk> i can send a mail to the list of all of the cherry picking done over the last month or so.
21:31:00 <danwent> garyk: yeah, that would be helpful
21:31:14 <garyk> danwent: ok, i'll do that tomorrow morning
21:31:26 <danwent> garyk: great.  thanks for driving the essex/stable stuff
21:31:56 <danwent> #topic community topics
21:32:10 <garyk> when do we star to work on gongys-1?
21:32:28 <danwent> ?
21:32:31 <gongys> ?
21:32:34 <danwent> :)
21:32:38 <garyk> the G version for OpenStack
21:32:42 <danwent> haha
21:32:45 <zhanghua> :D
21:32:56 <danwent> Summit is mid-october
21:33:26 <danwent> but the real answer is, once we finish up the critical items for Folsom, and that depends on how quickly we work :)
21:33:44 <danwent> usually after the final interation we sliip into a mode focused on testing, bug fixing and documenting.
21:34:02 <danwent> then once we're happy with the state of a release, we start thinking about the next summit.
21:34:14 <SumitNaiksatam> we need a poll to choose between garyk-1 and gongys-1 :-)
21:34:15 <danwent> the actual folsom release date is late september
21:34:20 <gongys> What a bad guy.
21:34:20 <garyk> ok, sounds good
21:34:27 <gongys> I like it.
21:34:28 <danwent> i still choose grizzly
21:34:53 <danwent> it would be unfair to pick either garyk or gongy, so grizzly is the clear choice :P
21:35:09 <danwent> also, a reminder to sign up for review days: http://wiki.openstack.org/ReviewDays
21:35:23 <danwent> some folks have been better at following their review days than others.
21:35:32 <garyk> is thursday a bug triage day?
21:35:45 <danwent> so far, i haven't written my review monitoring scripts though, so some of you get a pass, for now
21:36:07 <edgarmagana> :-0
21:36:27 <danwent> garyk: i heard ttx mention something about a bug triage day
21:36:33 <danwent> any know for sure
21:37:27 <danwent> #todo #danwent look into possible bug triage day
21:37:54 <danwent> to be honest, i'd probably rather wait until we finish F-3 and to then get a hold on our bugs right before the main bug-fixing period starts
21:37:58 <danwent> but we'll see.
21:38:17 <danwent> one last note: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
21:38:23 <danwent> this is the new openstack dev mailing list
21:38:47 <danwent> our plan has been to deprecate the netstack list and move to the main openstack-dev list, now that we're a core project.
21:38:59 <gongys> ok
21:39:11 <danwent> all emails related to quantum should have [Quantum] in the subject line
21:39:23 <danwent> hopefully that will make for easier scanning/filtering
21:39:52 <danwent> I personaly will probably have two different folders, one for main openstack-dev list and one for openstack-dev tagged with quantum
21:40:06 <danwent> as its pretty hard to keep up with the overall rate of email on the main openstack list
21:40:13 <ncode> danwent: awesome, it will make easy to follow quantum related messages
21:40:31 <danwent> the current openstack list will keep being used, but will be focused more on questions about the use of openstack (not dev discussions)
21:41:03 <danwent> #topics open discussion
21:41:17 <danwent> anything?  from anyone?
21:41:42 <gongys> any idea about my spec about floating ips?
21:41:52 <zhanghua> I'm going to dedicated to this project.
21:42:07 <danwent> zhanghua: dedicated to floating IPs?
21:42:30 <zhanghua> I mean, Quantum, network service
21:42:40 <danwent> zhanghua: ah, great.
21:42:53 <danwent> zhanghua: there's definitely plenty of work to do, so it will be really helpful
21:43:03 <danwent> gongys: sent you mail about this.
21:43:04 <zhanghua> I have seen a lot of progress here. amazing.
21:43:28 <gongys> I have a mail sent back.
21:43:33 <danwent> gongys: had planned to cover it under: https://blueprints.launchpad.net/quantum/+spec/quantum-l3-fwd-nat, but would be great to have help with this BP.  So would be great if you could drive the floating IPs side of things.
21:43:47 <danwent> gongys: ah, ok, am (as always) a bit behind on mail :)
21:44:13 <danwent> ok, anything else?
21:44:41 <gongys> Also I think we should have a spec for each BP before we start work.
21:44:44 <danwent> btw, i really want to put together some states on contributions to quantum in the past month or so.  I think the number of commits and number of participatns has really jumped, which is awesome
21:44:49 <gongys> This will help reviewing.
21:45:00 <danwent> gongys: ah, thanks for mentioning that… was on the agenda but i blew past it.
21:45:09 <danwent> I strongly agree.
21:45:26 <danwent> for any significant feature, we should require a spec before reviewing
21:45:42 <danwent> sometimes its really hard to get the big picture just by wandering through the code in a review
21:45:59 <gongys> Sometimes, I have to read codes for reviewing, to guess what is for, and how to use it, which is hard.
21:46:30 <zhhuabj> very good idea
21:46:34 <danwent> gongys: definitely.  I'm going to start pinging people for specs earlier in the development cycle as well
21:46:34 <ncode> +1
21:47:01 <danwent> if you have something priority high or above, expect to get emails if you don't have a spec up two weeks into the iteration.
21:47:36 <danwent> (this doesn't need to be a completely baked spec, but at least the key use cases, basic design, and touch points with other components)
21:47:54 <gongys> Yes, that is enough.
21:48:19 <danwent> ok, anything else folks?
21:48:22 <zhhuabj> if there is a template ?
21:49:01 <danwent> zhanghua: a spec template?  not at this point.  it might be helpful to develop a suggested one though.
21:49:42 <danwent> include sections like "use cases", "new apis", "ipmlementation components", "interactions with external components"
21:49:56 <zhhuabj> i see
21:50:09 <danwent> i dont' think we need to enforce one yet, but we can put together a suggested one on the wiki for anytone who wants guidence
21:50:19 <danwent> what do others think?
21:50:33 <mestery> A spec template seems like a really good idea.
21:50:43 <zhhuabj> dan, zhanghua and me (zhhuabj) are two people
21:50:50 <zhanghua> :D
21:51:09 <zhanghua> yes
21:51:13 <ncode> it's good and will help a lot new contributors
21:51:57 <danwent> #todo #danwent send out link to suggested spec template
21:52:05 <zhanghua> I agree with zhhuabj to have some spec guidance.
21:52:53 <danwent> ok, anything else folks?
21:53:02 <zhanghua> This will help developer to have a good thinking model.
21:53:42 <danwent> zhhuabj: hehe, sorry, auto-complete in my irc client :)
21:54:02 <danwent> got to pay more attention
21:54:12 <zhhuabj> :)
21:54:17 <zhanghua> :D
21:54:24 <danwent> ok, have a good week folks!
21:54:29 <danwent> #endmeeting