13:00:25 <joehuang> #startmeeting tricircle
13:00:26 <openstack> Meeting started Wed Dec 14 13:00:25 2016 UTC and is due to finish in 60 minutes.  The chair is joehuang. Information about MeetBot at http://wiki.debian.org/MeetBot.
13:00:26 <hejiawei> hi
13:00:27 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
13:00:29 <openstack> The meeting name has been set to 'tricircle'
13:00:41 <joehuang> #topic rollcall
13:00:53 <joehuang> #info joehuang
13:00:54 <zhiyuan> #info zhiyuan
13:01:00 <longxiongqiu> #info longxiongqiu
13:01:02 <Yipei_> #info Yipei
13:01:05 <hejiawei> #info jiawei
13:01:05 <RongHui11> #info ronghui
13:01:16 <liukun> #info liukun
13:01:33 <yinxiulin> #info xiulin
13:02:26 <joehuang> #topic VxLAN L2 networking across Neutron feature
13:02:55 <joehuang> currently the combined bridge network spec is approaching to be merged
13:03:14 <joehuang> but still want to wait for more review comment for one to two week
13:03:31 <joehuang> and the code is also submitted by zhiyuan for review too
13:03:38 <liuzeyu> #info liuzeyu
13:03:59 <joehuang> it's time for use to start the discussion of VxLAN L2 networking across Neutron
13:04:15 <joehuang> as you know, currently only shared VLAN is supported
13:05:06 <joehuang> Junsik told me he is not able to attend the meeting, but he sent me some info about his experiment on L2 networking across openstack cloud
13:05:31 <joehuang> your thoughts?
13:06:30 <zhiyuan> how did Junsik do the experiment? manually add the flow?
13:07:00 <joehuang> To connect Multi-region OpenStack instances, I add an additional bridge named "br-cap" and configure tunnel ports on the bridge. And the bridge "br-cap" is controlled by ONOS SDN controller.
13:07:04 <joehuang> To automate the configuration, I introduced a concept of template-based automation . I defined a self-defined template format to describe virtual network topology over multiple servers.
13:07:08 <joehuang> Then my tool automates the configuration of OVS bridges and VXLAN ports via OVSDB protocol, and calls ONOS APIs to down OpenFlow rules to "br-cap" bridges.
13:07:11 <joehuang> As the basic flow rules, we just set broadcast flow rules not fine-grained flow rules. For example, a packet arrives br-cap from br-int, then "br-cap" broadcast the packet to all tunnels.
13:07:39 <joehuang> the picture can not be pasted here
13:09:46 <joehuang> the idea is interesting
13:10:40 <zhiyuan> the ONOS SDN controller interact with Neutron server via ML2 mechanism driver?
13:11:17 <joehuang> I don't think so
13:11:51 <joehuang> we need to have a meeting to discuss this with Junsik
13:12:25 <Yipei_> so one port of the br-cap connects to br-int, the other connects to the br-tun?
13:12:53 <zhiyuan> i think br-tun is not needed
13:13:05 <zhiyuan> br-cap does the work
13:13:11 <joehuang> I'll try to find a place to put the pictures sent by Junsik
13:14:01 <joehuang> and during the barcelona summit,when we talked with Kevin about the L2 networking across Neutron, Kevin prefer to rk across Neutron
13:14:36 <Yipei_> is it ok for neutron community that we abandon br-tun?
13:15:39 <joehuang> and Kevin prefer to store the vtep in port binding info for push notification
13:15:51 <joehuang> and by the way for external communication
13:16:02 <joehuang> http://lists.openstack.org/pipermail/openstack-dev/2016-November/107061.html
13:18:28 <joehuang> It take times to think over the method for VxLAN l2 networking across Neutron
13:19:31 <joehuang> let's have a meeting at 10:00am (utc 2:00am), I'll provide an etherpad link tomorrow in IRC
13:19:49 <zhiyuan> ok
13:20:09 <joehuang> #action have a online meeting to discuss VxLAN l2 networking across Neutron
13:20:35 <Yipei_> tomorrow´╝č
13:20:42 <joehuang> #info on UTC2:00am  Dec 15
13:21:23 <joehuang> #info i.e Beijing time 10:00 am Dec.15, Korea/Japan 11:00 am Dec 15
13:21:45 <joehuang> #info etherpad link will be provided in IRC
13:22:14 <joehuang> #topic Ocata feature development review
13:22:28 <joehuang> let's have a short review on the feature development
13:22:55 <joehuang> please describe shortly and discuss as needed
13:24:36 <longxiongqiu> qos code has completed , i need some time to test in two nodes envir
13:24:48 <RongHui11> fix some environment patch and continue to perfect the configuration guide
13:24:51 <Yipei_> for lbaas, learn to write lbaas service plugin, and writing spec.
13:24:53 <zhiyuan> (1) patch to fix resource deletion has got one +2 (2) patch for bridge network combination is still under review, as Joe mentioned (3) a new patch to adpat DVR in cross-pod l3 networking has been submitted (4) start to consider async job reliability improvement
13:24:59 <joehuang> o, xiongqiu, your blueprint and spec also needed
13:25:30 <yinxiulin> I am doing network and subnet resources update
13:25:30 <joehuang> to Ronghui, good, configuration guide is better and better
13:25:46 <dongfeng> releasenote for routing api is delayed for there is a math test this week. and i will deal with it this weekend.
13:26:03 <joehuang> to Zhiyuan, great progress!
13:26:19 <joehuang> to Xiulin, keep going
13:26:40 <yinxiulin> ok, i will
13:26:47 <joehuang> to Dongfeng, it's ok for release notes, it's not very urgent
13:27:06 <dongfeng> ok.
13:27:30 <longxiongqiu> ok
13:27:57 <joehuang> to Yipei, good, for LBaaS, we may need some online discuss, could you organize it?
13:28:07 <joehuang> before you submit the spec
13:28:29 <Yipei_> ok
13:29:43 <hejiawei> for me, I am claening tables,but  when some tables removed,related futures should update.
13:29:53 <Yipei_> i will create an etherpad and write down some items to be discussed, when i have a better understanding of lbaas
13:30:34 <joehuang> to Jiawei, which feature needs to update, you mean the az_hint for external network?
13:32:09 <hejiawei> yes.
13:32:26 <joehuang> is there any other features need to update?
13:33:08 <dongfeng> pod binding api and its doc should be removed,and resource api shoud be changed too
13:33:28 <dongfeng> resource routing api shoud be changed too
13:34:01 <joehuang> for pod table, how do you think about to change the pod_name to region_name
13:34:22 <hejiawei> I just remove some tables that not need update future
13:34:55 <joehuang> it's ok to clean step by step
13:35:03 <joehuang> with several patches
13:35:18 <joehuang> small patch will get merged sooner
13:35:36 <joehuang> to Dongfeng, yes, these should be update too
13:35:41 <hejiawei> so,next week,maybe I will meet some trouble,I am going to ask you in irc
13:35:52 <joehuang> no problem, welcome
13:36:13 <joehuang> for pod_name, I searched the repository, seems 30 places found
13:36:35 <dongfeng> could i complete the pod binding api clean part?
13:36:59 <hejiawei> of course. :)
13:37:14 <dongfeng> including pod binding api clean, it's doc clean, and resource api update.
13:37:34 <joehuang> yes, dongfeng, please
13:37:43 <joehuang> it's open source
13:37:50 <joehuang> colloboration is welcome
13:37:55 <dongfeng> fine. i will deal with them later.
13:37:57 <RongHui11> it also need some synchronize from tricircle to trio2o and i am working on it
13:38:09 <joehuang> ok, Ronghui
13:38:43 <joehuang> #info smaller patch will get merged sooner
13:39:06 <joehuang> for az_hint, I have one idea
13:39:29 <joehuang> usually az_hint will be put with az name list
13:39:51 <joehuang> but for tricircle, we map az to region
13:40:10 <joehuang> and for external network, input region in az_hint
13:40:42 <joehuang> so when there is need to lookup the az, will search the pod first
13:41:01 <joehuang> and if the az name in the pod table is not matched,
13:41:24 <joehuang> we can search again whether the region_name can be matched
13:41:49 <joehuang> this will bring flexibility to network/router az_hint input
13:41:58 <joehuang> how do you think about this idea?
13:44:07 <zhiyuan> how about users specify az in the az_hint when creating external network?
13:45:29 <joehuang> first match AZ(it's still one parameter in pod table), if not find try to match pod_name
13:46:08 <zhiyuan> but we need pod for external network
13:46:29 <zhiyuan> az may contain more than one pod
13:46:30 <joehuang> if more than one, should we check to see if there is external network in that pod?
13:46:56 <hejiawei> zhiyuan's idea sounds easier
13:47:20 <zhiyuan> but at the beginning no external networks in pods
13:47:22 <joehuang> may be two search function
13:47:43 <joehuang> one find_by_region used for external network creation
13:47:54 <joehuang> the other one find_by_az
13:48:14 <joehuang> in find_by_az, first match with az_name, then try pod_name
13:48:49 <hejiawei> I don't understand these code about external network,so I should read them,I can just try your ider after I understand them
13:48:55 <hejiawei> to joe
13:49:06 <zhiyuan> ok, so for external network creation, users still need to specify pod_name, or say region_name
13:50:50 <joehuang> in fact, it's for router/network, the az_hint will be more flexible, you can specify az_name or region_name
13:51:28 <joehuang> for the enduser, specify region will be more intuitive when creating network
13:52:08 <zhiyuan> Another approach is to add "ready_for_external" network attribute to pod, then if az_name is specified when creating external network, we can filter pods by that attribute
13:52:58 <zhiyuan> sorry, shoule be "ready_for_external_network" attribute
13:52:59 <joehuang> yes, this way can keep the sematics of az_hint
13:53:28 <zhiyuan> and users can also use region_name if they like
13:53:49 <zhiyuan> so we don't have special handle for external network
13:53:59 <joehuang> good proposal
13:54:53 <joehuang> the parameter is too long, a shorter one will be better
13:54:59 <joehuang> :-)
13:55:32 <zhiyuan> sure
13:56:21 <joehuang> ok, let's have a discussion about L2 network tomorrow
13:56:43 <joehuang> any other topic
13:56:47 <zhiyuan> maybe a BP? the change is not hard, but many things like code, readme, api document need to be changed
13:56:55 <joehuang> yes
13:57:16 <joehuang> but this update will make the usage more friendly
13:57:35 <RongHui11> no for me
13:57:45 <zhiyuan> no other topic
13:57:48 <Yipei_> no
13:57:50 <dongfeng> no for me
13:57:52 <hejiawei> no for me
13:58:08 <yinxiulin> no
13:58:19 <joehuang> ok, thank you
13:58:22 <joehuang> #endmeeting