21:01:37 <markmcclain> #startmeeting Networking
21:01:38 <openstack> Meeting started Mon Jun 17 21:01:37 2013 UTC.  The chair is markmcclain. Information about MeetBot at http://wiki.debian.org/MeetBot.
21:01:39 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
21:01:41 <openstack> The meeting name has been set to 'networking'
21:01:44 <enikanorov> hi
21:02:00 <armax> hello
21:02:03 <danwent> hi
21:02:08 <markmcclain> #link https://wiki.openstack.org/wiki/Network/Meetings
21:02:43 <markmcclain> #topic Announcements
21:03:12 <danwent> hi folks, i sent a note to the core email alias already, but i'm stepping down as a member of quantum core
21:03:25 <mestery> danwent: Thanks for all your hard work over the years!
21:03:30 <danwent> don't have sufficient quantum review cycles with my new role
21:03:40 <danwent> but its been a fun time, and i really enjoyed working with you all :)
21:04:28 <marun> danwent: Thank you!
21:05:06 <danwent> mestery, marun: thanks
21:05:34 <armax> danwent: we sure will miss you!
21:05:54 <mlavalle> danwent: good luck! I hope your new role includes interacting with Rackspace
21:06:20 <marun> danwent: We all know how much blood, sweat and tears you put into this project over the years, and wouldn't have gotten here without you leading the charge.
21:06:22 <danwent> thanks.  sorry, don't mean to delay the meeting.  please proceed markmcclain :)
21:07:09 <markmcclain> danwent: No delay.  Your hard work spent leading this project is the reason the project is where it is today.
21:07:36 <danwent> ok, now this is just embarassing… let's move on :P
21:08:04 <markmcclain> Moving on to less sappy items… we have lots of inflight blueprints for H2
21:08:53 <markmcclain> July is rapidly approaching.. next week we'll spend some time talking about the critical and high priority blueprints for H2
21:09:23 <markmcclain> several of the blueprints are dependencies for later work
21:09:36 <markmcclain> with that lets run through the sub-teams
21:09:39 <markmcclain> #topic API
21:10:00 <markmcclain> salv-orlando: hi
21:10:09 <salv-orlando> aloha people
21:10:14 <salv-orlando> nothing major to repor
21:10:15 <salv-orlando> t
21:10:35 <salv-orlando> I think we are on track with the API related blueprints I am actively following.
21:10:50 <salv-orlando> We also have a 'provider-router' blueprint
21:10:55 <salv-orlando> that touches API and L3
21:11:06 <salv-orlando> whose spec is under review.
21:11:17 <salv-orlando> You're welcome to chime in.
21:11:23 <salv-orlando> Our priorities for H2 should be:
21:11:38 <salv-orlando> https://blueprints.launchpad.net/quantum/+spec/api-core-for-services
21:12:14 <salv-orlando> and quantum-l3-routing-plugin
21:12:22 <salv-orlando> https://blueprints.launchpad.net/quantum/+spec/quantum-l3-routing-plugin
21:12:24 <markmcclain> agreed
21:12:35 <salv-orlando> It would be nice also getting in that blueprint about router ports… can't find the link now
21:12:43 <salv-orlando> markmcclain: I mean your blueprint
21:13:14 <markmcclain> yeah.. needs another round of rebase to post
21:13:44 <salv-orlando> enikanorov is doing most of the practical work api-core-for-services - I am providing 'advice' (which you can translate as occasionally answering to queries on IRC)
21:13:48 <salv-orlando> that's all
21:13:58 <markmcclain> ok.. cool
21:14:12 <markmcclain> any API questions?
21:14:27 <markmcclain> #topic VPNaaS
21:14:33 <markmcclain> nati_ueno: hi
21:14:56 <nati_ueno> we can start playing with VPNaaS https://wiki.openstack.org/wiki/Quantum/VPNaaS/HowToInstall
21:15:04 <nati_ueno> however code is still early stage
21:15:33 <nati_ueno> There is three patch 1) Quantum Advanced Service Plugin for VPNaas (Patchset 9) https://review.openstack.org/#/c/29812/
21:15:41 <nati_ueno> 2) Quantum Client for VPNaaS (Patchset 6) https://review.openstack.org/#/c/29811/
21:15:50 <nati_ueno> 3) Agent Impl https://blueprints.launchpad.net/quantum/+spec/ipsec-vpn-reference (WIP: https://review.openstack.org/#/c/33148/1
21:16:19 <markmcclain> Thanks for the HowTo wiki
21:16:20 <nati_ueno> Main TODO is unit test and integration test. We need to be rebase to newer service directory structures also
21:16:29 <nati_ueno> that's all from me
21:16:52 <markmcclain> cool.. thanks for the update
21:17:04 <markmcclain> #topic Nova Integration
21:17:34 <markmcclain> garyk: awake :p
21:17:47 <garyk> markmcclain: yeah.
21:18:01 <garyk> markmcclain: there are actually a few issues. give me a sec to paste them
21:18:40 <garyk> markmcclain: some are related to reviews in nova:- security groups, refreshing of floaing iip's and passing host parameters
21:19:02 <garyk> 1. https://review.openstack.org/#/c/27709/ - do we support this with our security groups?
21:20:14 <markmcclain> we don't have anything comparable currently
21:20:36 <garyk> 2. https://review.openstack.org/#/c/33054/ - this was on the mail list too. updating a floating ip did not refresh nova's db
21:21:19 <garyk> 3. we need this for the multihost - https://review.openstack.org/#/c/29767/
21:21:39 <markmcclain> 2. solves part of the issue, but not the full problem
21:22:10 <markmcclain> if the IP is disassociated via the Quantum API then Nova's cache will be wrong
21:22:18 <danwent> garyk:  on #1, it seems like at some point, nova needs to stop taking network changes.
21:22:52 <markmcclain> danwent: I agree with you on #1
21:23:05 <garyk> danwent markmcclain : agreed.
21:23:11 <danwent> we should loop nova leadership in on that, and try to get their core devs aware of this as well.
21:23:24 <danwent> (or, at least that is what I would suggest :P)
21:23:32 <markmcclain> I'll follow up with Russell
21:23:38 <garyk> markmcclain: yes,s that is correct the device_id for the floating ip port is not the instance id
21:24:00 <garyk> markmcclain: i guess that is all at the moment.
21:24:11 <mestery> Looks like sdague agrees with everyone on that, he gave a -1 to that review with the same comments.
21:25:07 <markmcclain> mestery: good point..I'll still follow up to make everyone is on the same page
21:25:21 <markmcclain> Any questions on Nova Integration?
21:25:23 <garyk> markmcclain: mestery: i think that is is importnat for us to try and show that we have the support in quantum then it can help move things forwards
21:25:42 <garyk> markmcclain: btw are we going to try and move quantum to the default with devstack?
21:25:50 <mestery> garyk: Agree, though Sean's point about ensuring no new features in nova-networking seems like a good idea to me.
21:25:54 <markmcclain> garyk: yes
21:26:09 <mlavalle> markmcclain: there is a whole list of nova / quantum incompatibilities I have discovered fixing tempest tests
21:26:20 <mlavalle> I will send a list to the ml
21:27:19 <markmcclain> mlavalle: sounds good
21:27:28 <markmcclain> #topic FWaaS
21:27:34 <markmcclain> SumitNaiksatam: hi
21:27:37 <SumitNaiksatam> Hi!
21:27:45 <SumitNaiksatam> We made progress on all the blueprints in the past week.
21:27:54 <SumitNaiksatam> CLI patch has been posted by KC: https://review.openstack.org/#/c/33187/
21:28:09 <SumitNaiksatam> this is WIP, but has been written to the current state of the API patch
21:28:36 <SumitNaiksatam> Sridar/Dan are also getting close to posting a WIP patch for the agent (our target was today initially)
21:29:01 <SumitNaiksatam> Rajesh has is working on the unit tests for the driver, he has some dependency on the agent code to develop the tests
21:29:10 <SumitNaiksatam> so he is a little blocked
21:29:16 <SumitNaiksatam> we are meeting tomorrow to get past this
21:29:28 <markmcclain> meeting in person or virtually?
21:29:36 <SumitNaiksatam> person
21:29:40 <markmcclain> ah ok
21:29:44 <SumitNaiksatam> we have been meeting regularly
21:29:51 <markmcclain> cool
21:29:55 <SumitNaiksatam> I am working on addressing the review comments on the API patch, and also finishing the unit tests
21:29:58 <SumitNaiksatam> but not done
21:30:27 <SumitNaiksatam> devstack patch should be in soon after the agent patch is posted (of course will be WIP)
21:30:37 <markmcclain> ok
21:31:02 <SumitNaiksatam> SridarK, dan_florea RajeshMohan anything more to add?
21:31:10 <RajeshMohan> Testing with quantum-router (namespace + iptables) is pending. Other than that, most of the manual testing is complete. I simulate Agent but everything else is real. Good for manual testing but not good for automation.
21:31:24 <dan_florea> Nothing beyond what you already said. We should have a WIP patch for agent this week.
21:31:24 <RajeshMohan> For automation, I have to mock iptables-manager. It will be easy to automate verification with Mock. Once that is done, I will publish unit-testcase related files for review.
21:31:51 <SumitNaiksatam> thanks RajeshMohan dan_florea
21:32:06 <markmcclain> Thanks all for the update
21:32:18 <markmcclain> #topic LBaaS
21:32:22 <markmcclain> enikanorov: around?
21:32:25 <enikanorov> hi
21:32:51 <enikanorov> lbaas got much attantion from cores recently. Thanks for that
21:33:02 <enikanorov> multivendor support step1 got in
21:33:12 <enikanorov> also some major renaming happened
21:33:24 <markmcclain> file renaming?
21:33:39 <enikanorov> directory renaming and moving services one level up
21:33:49 <markmcclain> ok
21:34:05 <enikanorov> this patch fixes devstack: https://review.openstack.org/#/c/32964/
21:34:34 <enikanorov> also there is a patch that adds agent-scheduling to the reference implementation: https://review.openstack.org/#/c/32137/4
21:35:15 <enikanorov> another feature frequently requested is to add multiple vips per pool
21:35:23 <enikanorov> I've started working on this.
21:35:34 <markmcclain> yeah.. I've seen a few threads
21:35:39 <markmcclain> about that it
21:36:04 <markmcclain> before committing to code… are you going to create a blueprint for how this will work?
21:36:29 <enikanorov> sure. BP is created and currently I'm experimenting with reference impl to make it work
21:36:41 <enikanorov> there VIPs are sharing single quantum port
21:36:46 <enikanorov> (and pool, of course)
21:37:21 <markmcclain> ok.. I missed that blueprint.. I'll find it after the meeting
21:37:23 <enikanorov> I'll post spec once I clear the details
21:37:33 <markmcclain> Anything else for LBaas?
21:37:36 <enikanorov> https://blueprints.launchpad.net/quantum/+spec/lbaas-multiple-vips-per-pool
21:37:37 <lifeless> enikanorov: why does devstack need fixing?
21:37:43 <enikanorov> nope, thanks.
21:37:48 <markmcclain> enikanorov: thanks!
21:37:49 <lifeless> enikanorov: looks like a backwards incompatible change to me?
21:37:53 <enikanorov> lifeless: directory renaming. need to fix plugin path
21:38:37 <lifeless> enikanorov: we -really- need to stop gratuitous backwards incompatible changes; everytime devstack needs fixing, we've just broken all trunk deployers, all CI systems running tests etc.
21:38:44 <markmcclain> lifeless: LBaaS in Grizzly was experimental
21:38:51 <lifeless> enikanorov: surely there is a backwards compatible way of doing the change ?
21:39:35 <lifeless> markmcclain: is it still experimental? If so, sure, take my comments as wishful thinking - its my head right now because of https://bugs.launchpad.net/tripleo/+bug/1184484/comments/12
21:39:38 <uvirtbot> Launchpad bug 1184484 in tripleo "Quantum default settings will cause deadlocks due to overflow of sqlalchemy_pool" [Critical,Triaged]
21:39:40 <enikanorov> lifeless: I don't think so. That was a major change that I believe should have been done.
21:39:59 <enikanorov> by the way
21:40:33 <enikanorov> As I've posted on VPN and FW reviews, I'd suggest not name the directories as 'agent_vpn' or similar
21:40:45 <markmcclain> ok
21:40:46 <enikanorov> as this points to particular plugin-agent architecture
21:41:31 <nati_ueno> enikanorov: We will change the namings
21:41:32 <markmcclain> lifeless: I understand your concern about backwards compatible changes
21:41:37 <enikanorov> and not a generic one. This would help to avoid bw incompatible renaming
21:41:41 <SumitNaiksatam> enikanorov: point well taken, we were not doing that in FW
21:42:10 <markmcclain> ok.. need to move onto ML2
21:42:12 <markmcclain> #topic ML@
21:42:16 <markmcclain> #topic ML2
21:42:22 <rkukura> good progress by the Modular L2 sub-team this past week!
21:42:35 <rkukura> mestery organized the 1st weekly IRC meeting
21:42:51 <rkukura> these will be Wednesday at 1400 UTC in #openstack-meeting going forward
21:43:20 <rkukura> initial code in review on two H-2 BPs:
21:43:27 <rkukura> https://blueprints.launchpad.net/quantum/+spec/ml2-mechanism-drivers code under review at https://review.openstack.org/#/c/33201/
21:43:36 <rkukura> https://blueprints.launchpad.net/quantum/+spec/ml2-gre code under review at https://review.openstack.org/#/c/33297/
21:43:47 <gongysh> hi, I am here at last.
21:43:57 <mestery> I should note the ml2-vxlan blueprint is somewhat dependent on the ml2-gre blueprint since the tunnel RPC calls are implemented in the former.
21:44:44 <markmcclain> mestery: I'll make sure they're marked as dependent
21:44:57 <markmcclain> gongysh: hi
21:45:00 <mestery> markmcclain: Thanks!
21:45:16 <gongysh> markmcclain: hi
21:45:29 <mestery> There is also this change which allows multiple tunnel_types in the OVS agent: https://review.openstack.org/#/c/33107/
21:45:37 <mestery> I'm reworking it a bit, hope to post a new version tomorrow.
21:45:49 <rkukura> mestery: cool - forgot to mention that one
21:46:22 <markmcclain> ok.. any other updates for ML2?
21:46:33 <rkukura> others are welcome to contribute, participate in the meetings, etc.
21:47:07 <markmcclain> Wednesdays at 1400UTC in #openstack-meeting
21:47:12 <rkukura> at some point we should require new functionality in ml2 at the same time (or instead of) the legacy plugins
21:47:30 <markmcclain> #topic clientlib
21:47:45 <mestery> rkukura: I think once the ml2-gre goes in it may make sense.
21:47:52 <mestery> Isn't that the last major gap (GRE tunneling support)?
21:48:14 <rkukura> yes, but we'll continue playing catch up
21:48:30 <markmcclain> rkukura: agreed… once we have parity we'll discuss deprecating the OVS and lb plugins
21:48:47 <markmcclain> and how the process will work so that we can communicate to everyone
21:48:57 <rkukura> agreed
21:49:20 <markmcclain> Moving onto the client library
21:49:34 <mestery> Have to take off now folks, later!
21:49:47 <markmcclain> mestery: later
21:49:51 <markmcclain> we've got a new alpha tarball that fixes the issues caused by pbr
21:50:03 <markmcclain> Please test this one: http://tarballs.openstack.org/python-quantumclient/python-quantumclient-2.2.2a1.tar.gz
21:50:35 <markmcclain> Took a little longer to tag a new version, but we had to wait until the upsteam dependencies were addressed first
21:51:48 <markmcclain> If you have some spare cycles to test, please do so and report back to the list.
21:52:08 <markmcclain> #topic Stable
21:52:33 <markmcclain> garyk: looks like we're all caught up on Stable reviews
21:53:21 <garyk> markmcclain: reviews are ok at the moment. i need to backport one cisco fix that i am battling with. i'll try and get mestery to help with this tomorrow :)
21:53:29 <markmcclain> ok
21:53:49 <markmcclain> #topic testing
21:54:10 <markmcclain> mlavalle: you've got some consistency bugs you're going to send out?
21:54:22 <mlavalle> yes I will
21:54:55 <markmcclain> ok
21:55:17 <markmcclain> thanks for keeping us in the loop
21:55:39 <markmcclain> #topic Other team reports
21:56:06 <markmcclain> I know the PXE review has stalled out a bit
21:56:21 <dkehn> yes, did get items on the client side
21:56:30 <dkehn> and working on testing end-to-end
21:56:36 <markmcclain> cool
21:56:40 <dkehn> and that about it
21:56:48 <dkehn> would like to push this through
21:57:07 <markmcclain> amotoki and I are reviewing the server side patch.. hopefully can get it through soon
21:57:16 <dkehn> sweet
21:57:27 <markmcclain> Any other team reports?
21:57:59 <markmcclain> #topic Open Discussion
21:58:23 <markmcclain> For those that missed it: http://www.openstack.org/summit/openstack-summit-hong-kong-2013/
21:58:37 <markmcclain> The links to the Hong Kong summit hotels have been posted
21:59:00 <markmcclain> additionally ATC codes for summit registration have started to be sent out
21:59:12 <markmcclain> so if you plan on going watch your email
21:59:17 <lifeless> hi
21:59:23 <lifeless> question about backwards compat
21:59:30 <lifeless> in this comment https://bugs.launchpad.net/tripleo/+bug/1184484/comments/12
21:59:31 <uvirtbot> Launchpad bug 1184484 in tripleo "Quantum default settings will cause deadlocks due to overflow of sqlalchemy_pool" [Critical,Triaged]
21:59:33 <markmcclain> lifeless: sure
21:59:57 <lifeless> salv-orlando says that [DATABASE] isn't valid anymore, I'm curious whether this is something in trunk, or in the two patches suggested to fix the bug.
22:00:16 <salv-orlando> in one of the two patches
22:00:22 <salv-orlando> which is still under review
22:00:34 <salv-orlando> most likely something needs to be fixed there before merging.
22:00:47 <lifeless> ah, ok, cool.
22:00:48 <salv-orlando> we haven't (yet) broken compatibility in trunk
22:00:55 <lifeless> I was going to be grumpy if you had :>
22:01:07 <lifeless> thanks!
22:01:12 <garyk> salv-orlando: lifeless: we are moving to oslo support - https://review.openstack.org/#/c/27265/
22:02:29 <gongysh> garyk: a bit patch.
22:02:35 <gongysh> a big patch
22:03:09 <markmcclain> gongysh: most of it is code from Oslo
22:03:15 <garyk> gongysh: yup, hopefully we can get it in sooner than later
22:03:27 <markmcclain> garyk: the olso config needs to land first correct?
22:03:34 <lifeless> garyk: so - right now that isn't backwards compat?
22:03:38 <salv-orlando> garyk: basically the issue we're talking about and is that the patch you linked ignores sql_connection
22:03:50 <garyk> markmcclain: correct
22:03:59 <salv-orlando> which should instead be used, as it should be the deprecated version of connection
22:04:21 <garyk> salv-orlando: lifeless: yes. we will need to make a minor change to ensure thatit is backward compatible
22:05:06 <lifeless> garyk: salv-orlando: does it still honour [DATABASE] at the moment? e.g. for testing if I just add a connection= line, it will be enough ?
22:05:49 <garyk> lifeless: yes, i think so
22:05:56 <lifeless> cool
22:06:03 <lifeless> salv-orlando: I will try the deadlock fix again today
22:06:09 <salv-orlando> k
22:06:49 <garyk> i am sorry i am going to call it a night - it is a few hours after bed time. good night
22:06:53 <markmcclain> Anything else?
22:06:55 <markmcclain> garyk: night
22:07:14 <markmcclain> Everyone have good week.
22:07:15 <markmcclain> #endmeeting