18:00:27 <bh526r_> #startmeeting gluon
18:00:28 <openstack> Meeting started Wed Feb  8 18:00:27 2017 UTC and is due to finish in 60 minutes.  The chair is bh526r_. Information about MeetBot at http://wiki.debian.org/MeetBot.
18:00:29 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
18:00:31 <openstack> The meeting name has been set to 'gluon'
18:00:42 <bh526r_> #topic Roll Call
18:00:48 <bh526r_> #info Bin Hu
18:00:57 <krenczewski> #info Kamil Renczewski
18:01:21 <jinli> #info JinLi
18:01:23 <jinli> Hi all
18:01:32 <krenczewski> Hi
18:01:40 <pcarver> #info Paul Carver
18:01:50 <bh526r_> Hi all
18:02:27 <bh526r_> #topic F2F Update
18:02:57 <bh526r_> #info We had an F2F on Monday Feb 6 and Tuesday 7 in Sunnyvale
18:03:35 <bh526r_> #info Our primary goal is to finalize Ocata deliverables and next steps
18:04:01 <georgk> #info Georg Kunz
18:04:05 <bh526r_> #info Regarding Ocata deliverables, we reviewed the task list
18:04:23 <bh526r_> #link https://wiki.openstack.org/wiki/Gluon/Tasks-Ocata
18:05:10 <bh526r_> #info The deliverables are those completed features.
18:05:30 <bh526r_> #info And documentation / installation guide
18:05:45 <bh526r_> #info for manual installation
18:06:15 <bh526r_> #info Regarding next steps, there have been comprehensive discussions
18:06:26 <bh526r_> #info The agreement is:
18:06:53 <alraddarla> #info Darla Ahlert
18:07:03 <bh526r_> #info 1. Extended ML2 Plugin for Gluon is the best way to integrate with Neutron
18:07:20 <bh526r_> #info So we will keep this direction
18:07:56 <bh526r_> #info 2. We can evolve Protons to be Neutron's Service Plugins. Some work to be done in the future
18:08:54 <bh526r_> #info In this case, i.e. Protons as Neutron's service plugin, we can reuse Neutron's DB instead of having our own MySQL DB
18:09:20 <bh526r_> #info Of course, development work is independently in Gluon Repo
18:10:35 <bh526r_> #info At certain time point, we will make Gluon to be part of Big Tent , and make Extended ML2 Plugin for Gluon and Proton Service Plugins to be part of Neutron Stadium and get full recognition of OpenStack community
18:11:08 <bh526r_> #info Ian and Sukhdev agreed to go to PTG in Atlanta and participate in discusson with Neutron and Nova
18:11:46 <bh526r_> #info Two specific items will be discussed: (1) Removal of “port” dependency on “network/subnet” (2) Network “type” attribute in Neutron
18:12:44 <bh526r_> Paul, we also discussed your point in last IRC meeting, i.e. bind port to 2 SDN controllers, similar to function of HPB in Neutron
18:13:06 <pcarver> ok
18:13:18 <bh526r_> Basically, Gluon/Proton is not an orchestrator
18:14:09 <bh526r_> An orchestrator will organize relevant data objects as parameters to call Proton APIs.
18:14:32 <bh526r_> Proton simply pass through data objects from Orchstrator to multiple SDN controllers
18:15:17 <bh526r_> When binding or other operation is requested, SDN controllers consume the data objects, constrained by policies, and process the binding and other operations
18:15:47 <bh526r_> All SDN controllers can monitor etcd, and handle the same port binding operation. But only the one that attaches to the VM can handle the virtual interface.
18:16:40 <bh526r_> It is up to the data objects from Orchestrator to specify one or multiple entities so that one Gluon port could be potentially bound to multiple SDN controllers, e.g. one SDN controller controlling VFs, and the other SDN controller controlling PFs.
18:17:07 <bh526r_> So it is the same model as HPB
18:18:46 <bh526r_> #info We also reviewed otustanding patches, and new patches from Kamal in terms of examples of policy.json
18:19:02 <bh526r_> #info all patches were approved
18:19:32 <bh526r_> Jin - one patch had verified-1, could you give some hints?
18:20:01 <jinli> Hi Bin, is it the one you email me?
18:20:18 <bh526r_> Yes
18:20:35 <jinli> I am checking it now
18:20:47 <jinli> will let you know by the end of day
18:21:25 <bh526r_> For 430450, you already rechecked it
18:22:15 <bh526r_> Let me know what happened then. Thank you Jin
18:22:26 <jinli> 430450 works
18:22:53 <bh526r_> It works now, but failed gate jobs yesterday with error messages that I sent you in the email
18:23:15 <jinli> right, will double check
18:23:39 <jinli> the test_as_dict failure cause by one of my test case
18:24:03 <bh526r_> Thank you.
18:24:58 <bh526r_> I see. Also my local tox failed with py35 after I pull most recent repo
18:25:14 <bh526r_> details in my email.
18:25:35 <bh526r_> Let me know after you look into it today
18:25:38 <jinli> ok
18:26:13 <bh526r_> Thank you
18:27:01 <jinli> NP
18:27:41 <bh526r_> #topic Ocata Task Update
18:28:11 <bh526r_> #link https://wiki.openstack.org/wiki/Gluon/Tasks-Ocata
18:28:25 <bh526r_> #info All completed tasks will be part of deliverables
18:29:50 <bh526r_> #info Synchronization operation etc. will be deferred to later (Ocata SR or Pike), because Ian needs more time to think through the code, and Sukhdev also indicated scalability consideration
18:30:50 <bh526r_> #info Installer support will be deferred too because it needs more time to complete relevant modules in different installers
18:31:13 <bh526r_> #info Manual Installation Guide, and some documentation will be provided.
18:31:46 <bh526r_> #topic Patch Status Update
18:32:01 <bh526r_> #link https://review.openstack.org/#/c/419210/
18:32:16 <bh526r_> #link https://review.openstack.org/#/c/422338/
18:32:32 <bh526r_> #link https://review.openstack.org/#/c/430443/
18:32:50 <bh526r_> #link https://review.openstack.org/#/c/430450/
18:33:03 <bh526r_> #info All are approved and merged
18:34:12 <bh526r_> Kamil do you feel the installation guide in CONTRAIL.md in your patch https://review.openstack.org/#/c/402071/ good enough for manual install Contrail Mechanism Driver?
18:34:53 <krenczewski> I think yes
18:35:34 <krenczewski> But it assumes that Contrail is installed
18:35:46 <krenczewski> I'll review it once more
18:36:08 <krenczewski> because I didn't touch this for some time since I started to work on fuel and devstack plugin
18:36:35 <bh526r_> Good. This is because when I read it, the first part of "Plugin" only says: Plugin ------ Plugin code is available here: https://github.com/codilime/ContrailMechanismDriver
18:36:57 <bh526r_> Then it jumped to Dependencies
18:37:14 <krenczewski> that is because this repository contrain it's own readme file
18:37:52 <krenczewski> and I didn't wanted to maintain two separate manual instructions
18:37:54 <bh526r_> I see, the we need to follow the instructions in your own repo
18:38:07 <bh526r_> That's right.
18:38:23 <krenczewski> I think this is the best way
18:39:14 <krenczewski> I'll try to emphasize this and re-submit a patch
18:39:20 <bh526r_> Perhaps adding one sentence at the end of "Plugin" section, indicating "Please follow the instruction in that repo to install Contrail Mechanism Driver" or something like that
18:39:43 <bh526r_> Then it will be crystal clear to users.
18:40:03 <bh526r_> Thanks
18:40:06 <krenczewski> right
18:40:41 <bh526r_> That's all from my side.
18:40:53 <bh526r_> Anything else you want to discuss?
18:41:30 <krenczewski> I think that is all
18:41:37 <bh526r_> Thank you Kamil
18:41:41 <krenczewski> Thanks
18:41:54 <bh526r_> Anyone else has AOB?
18:42:24 <bh526r_> If no, we can give back everyone 18 minutes
18:42:32 <bh526r_> #info Meeting Adjourned
18:42:44 <bh526r_> #endmeeting