09:00:23 <oanson> #startmeeting Dragonflow
09:00:24 <openstack> Meeting started Mon Sep 26 09:00:23 2016 UTC and is due to finish in 60 minutes.  The chair is oanson. Information about MeetBot at http://wiki.debian.org/MeetBot.
09:00:25 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
09:00:27 <openstack> The meeting name has been set to 'dragonflow'
09:00:32 <oanson> Hello.
09:00:41 <oanson> Who is here for the Dragonflow weekly meeting?
09:00:49 <hujie> hello
09:00:51 <yuli_s> I am
09:00:54 <xiaohhui> hi
09:01:00 <lihi> Hi
09:01:36 <oanson> Let's wait a minute. Maybe nick-ma and jingting will join us. Then we can start
09:02:18 <oanson> All right.
09:02:30 <oanson> For a change of pace, we have an Agenda today :)
09:02:34 <oanson> #topic Roadmap
09:02:46 <oanson> Newton is coming to an end very soon.
09:03:30 <oanson> So features to be added may have to wait for Ocata. Depends on the feature, its importance, impact, and risk.
09:04:03 <oanson> If there is something that looks like its being postponed but you think should be pushed, let us know.
09:04:40 <yuli_s> i thin redis patch should be implemented
09:04:41 <oanson> VLAN and Flat networks look like they are very close to the finish line. I would like to merge them. There is the small testing issue left.
09:04:52 <oanson> xiaohhui, this can be closed this week, right?
09:05:10 <xiaohhui> I think so, I don't know there still have test isue.
09:05:40 <xiaohhui> It has been resolved, unless there is new issue
09:05:48 <oanson> Sorry, I see it is solved.
09:05:58 <oanson> I will test and approve the changes this week.
09:06:14 <xiaohhui> Thanks~
09:06:32 <oanson> Same goes for Remote Port Function.
09:06:46 <hujie> yes Omer
09:06:50 <oanson> I see that nick-ma and xiaohhui already approved it
09:07:01 <oanson> I will go over it (hopefully today) and it can be merged as well.
09:07:10 <hujie> ok thx
09:07:18 <oanson> Port status notification:
09:07:44 <oanson> I approved it. Waiting for nick-ma to approve it as well.
09:08:07 <oanson> The patch is https://review.openstack.org/#/c/337659
09:08:46 <oanson> I see that gampel and nick-ma also approved https://review.openstack.org/#/c/323057/ (Extra routes support)
09:09:00 <oanson> I am not sure why they didn't merge it, but I think I can merge it today too.
09:09:15 <xiaohhui> What about the qos nb patch? https://review.openstack.org/#/c/331932/ Should we wait for O release?
09:10:00 <oanson> If possible, I think that's best
09:10:24 <oanson> It's only a few weeks away, and  I want to concentrate a bit on bug closing.
09:10:32 <xiaohhui> OK
09:10:47 <oanson> This will also give us time to add northbound and southbound in the same release cycle.
09:10:59 <oanson> And we can do it while not in a rush.
09:11:21 <oanson> If it ends up being really important for Newton, we can always backport it later.
09:11:24 <xiaohhui> Sounds reasonable
09:11:43 <oanson> Anything else for roadmap?
09:12:11 <xiaohhui> https://review.openstack.org/#/c/339975/
09:12:19 <xiaohhui> And this? Active port
09:12:55 <xiaohhui> But it is also not all code for that functionality, maybe we should also wait for O
09:13:16 <oanson> I would prefer to wait. It is a very big patch across many files.
09:13:37 <oanson> I saw the wangjian split it to smaller patches, but it is still very big.
09:14:20 <xiaohhui> It is actually yuanwei :)
09:15:03 <oanson> Sorry, yuanwei :)
09:15:08 <oanson> Read the wrong field on the patch :)
09:16:02 <oanson> Any other items here?
09:16:15 <xiaohhui> Not from me
09:16:21 <oanson> Very good.
09:16:26 <oanson> #topic Performance
09:16:33 <oanson> yuli_s, you weren't here last week.
09:16:42 <oanson> I reported about the performance results we have found then.
09:16:42 <yuli_s> yup,
09:16:48 <yuli_s> great !
09:16:53 <oanson> Is there anything new you want to sure?
09:16:54 <oanson> New results?
09:16:58 <oanson> Future test plans?
09:17:04 <yuli_s> sure,
09:17:17 <oanson> Additionally, if there are tests anyone here wants us to run, please speak up as well.
09:17:33 <yuli_s> i fixed one performance issue with cleaning openflow dhcp related rules
09:17:41 <yuli_s> this patch got merged,
09:18:13 <oanson> Great!
09:18:24 <yuli_s> i was researching ways to improve the time takes to create a group of subnets
09:18:55 <yuli_s> it takes around 50 seconds to create 50 subnets and link to rounter
09:19:12 <yuli_s> in regular neutron installation it is also not very fast
09:19:28 <yuli_s> it can take up to half a second to create a subnet
09:19:45 <yuli_s> i am going to run a test
09:19:55 <oanson> All right.
09:20:01 <yuli_s> to run 4k df servers
09:20:18 <yuli_s> and see how much time it takes to sync a new server
09:20:34 <oanson> When you say it takes 1/2 a second to add a subnet - this is in Dragonflow, or Neutron's reference implementation?
09:20:50 <oanson> yuli_s, that's an interesting test.
09:21:02 <yuli_s> default neutron installation
09:21:09 <yuli_s> without dragonflow
09:21:21 <oanson> So Neutron manage it in half the time we do?
09:21:31 <yuli_s> yes,
09:21:40 <oanson> I see.
09:22:10 <oanson> This will probably have to be fixed.
09:22:37 <xiaohhui> Yeah, we should take at least the same time as neutron do
09:22:40 <yuli_s> i will research this topic
09:22:46 <yuli_s> and look for ways fro improvements
09:22:55 <oanson> Did you also test port creation times for Neutron?
09:23:14 <yuli_s> nope
09:23:23 <yuli_s> not this time
09:23:23 <oanson> That might also be interesting.
09:23:40 <oanson> All right.
09:24:06 <oanson> If there are tests anyone here thinks we should run, please speak up
09:24:15 <xiaohhui> Are there any test about ovn? Maybe we should also compare with ovn.
09:24:50 <xiaohhui> And I am sorry, I don't remember if oanson has mentioned last week. Are the test run with ml2 or DFplugin?
09:25:15 <oanson> Actually, this is an important point.
09:25:28 <oanson> If the tests were run with DFPlugin, they should be run with ML2 as well.
09:25:41 <oanson> ML2 should become our standard configuration.
09:25:51 <xiaohhui> I think ml2 is more important, if that is true
09:26:10 <oanson> yuli_s, ?
09:26:25 <yuli_s> i will add this to my list
09:26:40 <oanson> Great, thanks!
09:27:43 <oanson> Anything else here?
09:27:48 <yuli_s> nop
09:27:53 <oanson> #topic Bugs
09:28:31 <oanson> I see there are a few new bugs that are undecided. I'll go over them and rank them.
09:29:02 <oanson> There are also some high bugs that have been solved. I'll mark them as such too.
09:29:28 <oanson> If possible, we should reach the end of Newton with 0 high bugs, but I am not sure this is possible.
09:29:57 <xiaohhui> I own 2 high priority bugs, I will try to resolve them
09:30:13 <oanson> All right.
09:30:14 <yuli_s> we have this bug to take care: 1624826
09:30:37 <nick-ma> i am in a meeting, sorry for late.
09:30:40 <oanson> I see feipeng has another 2. One is in progress, I don't know if he saw the other one.
09:30:48 <oanson> nick-ma, hi.
09:31:20 <xiaohhui> yuli_s: actually, I am working on bug 1624826
09:31:21 <openstack> bug 1624826 in DragonFlow "Openflow error message is hard to understand" [Medium,New] https://launchpad.net/bugs/1624826
09:31:44 <xiaohhui> https://review.openstack.org/#/c/375781/
09:32:02 <yuli_s> xiaohhui, great
09:32:11 <oanson> Yes. Definitely!
09:33:03 <xiaohhui> I had discussion with feipeng, I think one bug he owns is invalid for now. But I don't know the bug number for now.
09:33:11 <oanson> All right.
09:33:24 <oanson> I think I'll go over the bug list later this week, and contact people off line.
09:33:29 <xiaohhui> Let me find it out, 1 min
09:33:33 <oanson> Maybe next week have a summary
09:34:05 <oanson> xiaohhui, I think it can wait. I don't want to hold people here for longer than necessary :)
09:34:22 <xiaohhui> https://bugs.launchpad.net/dragonflow/+bug/1621451
09:34:23 <openstack> Launchpad bug 1621451 in DragonFlow "Errors occur when run dragonflow with redis pub/sub enabled" [High,New] - Assigned to feipeng (feipeng1)
09:34:24 <xiaohhui> This one
09:35:04 <oanson> I see.
09:35:22 <oanson> If feipeng (or you) could provide a link to the relevant commit, the bug could be closed.
09:35:34 <xiaohhui> Sure
09:35:38 <oanson> Thank you.
09:36:08 <oanson> As I said, I'll go over the list, and contact people off line if needed.
09:36:15 <oanson> Next week I'll have a summary ready.
09:36:20 <oanson> #topic Open Discussion
09:36:23 <xiaohhui> Great
09:36:27 <oanson> The floor is for the taking.
09:36:28 <nick-ma> ok
09:37:18 <xiaohhui> I don't know if you guys noticed, but there is error log in the ml2 fullstack, although the fullstack report success.
09:37:20 <xiaohhui> http://logs.openstack.org/59/376159/1/check/gate-dragonflow-dsvm-fullstack-ml2-nv/9cc9ef0/logs/screen-q-svc.txt.gz?level=ERROR
09:38:35 <oanson> This looks important. It may mean that ports aren't being updated in the NB database.
09:38:54 <xiaohhui> I found it when try to identify some other fullstack issue.
09:39:02 <nick-ma> i wonder why the gate is success.
09:39:23 <oanson> I think update_port_postcommit in Neutron is called with continue_on_failure=True.
09:39:37 <oanson> Maybe the update is small enough that we don't test what's changed.
09:40:22 <oanson> I'll open a bug.
09:40:36 <nick-ma> with high priority. i think.
09:40:41 <oanson> nick-ma, yes.
09:40:48 <xiaohhui> I would take a deep look into it, if no one else object it.
09:40:54 <oanson> If anyone is available, please take it. Otherwise I will have to look at it, but only at the beginning of next week.
09:41:04 <oanson> xiaohhui, it's yours! :)
09:41:09 <xiaohhui> :)
09:41:16 <nick-ma> thanks.
09:41:44 <oanson> Floor is free again.
09:42:31 <oanson> All right.
09:42:39 <oanson> If there is nothing else.
09:42:44 <oanson> Thanks everyone for coming.
09:42:53 <oanson> And thanks everyone for your effort and good work!
09:42:57 <nick-ma> thanks all.
09:43:09 <oanson> #endmeeting