#openstack-meeting: gluon

Meeting started by bh526r at 18:01:09 UTC (full logs).

Meeting summary

  1. Roll Call and Introduction (bh526r, 18:02:32)
    1. Georg Kunz (georgk, 18:02:55)
    2. Kamil Renczewski (KamilRenczewski, 18:03:01)
    3. Bin Hu (bh526r, 18:03:03)

  2. Admin Update (bh526r, 18:03:31)
    1. we plan to have a Gluon F2F session in Barcelona (bh526r, 18:04:17)
    2. Currently OpenStack is allocating rooms for official projects, which is the priority (bh526r, 18:04:57)
    3. Bin communicated with Thierry, and agreed that once the room assignment for official projects has been done, if there is more spaces available, we can get it (bh526r, 18:05:59)
    4. Bin Hu Right now, the space in Barcelona is tighter compared to prior summits (bh526r, 18:06:39)
    5. Paul Carver (pcarver, 18:06:41)
    6. Bin will also try to ask for a slot in Nova and/or Neutron's sessions in Barcelona to introduce Gluon (bh526r, 18:09:06)

  3. Summary of Face-to-Face Meeting on August 18 and 19 (bh526r, 18:10:06)
    1. Gluon had an F2F meeting in Sunnyvale on 8/18 and 8/19. (bh526r, 18:10:44)
    2. The summary of discussion and conclusion is as follows: (bh526r, 18:11:05)
    3. 1. Group discussed Gluon and Neutron Interaction Model. Group agreed and concluded that: (bh526r, 18:12:23)
    4. 1.1 A Proton stays the same. i.e. YAML -> { Pecan, DB Schema, etcd interface } (bh526r, 18:13:05)
    5. 1.2 Gluon will use Neutron plugin approach for North Bound, i.e. as a Gluon ML2 driver (core plugin) (bh526r, 18:13:50)
    6. - Gluon Wrapper ML2 Driver subclasses ML2 Driver (bh526r, 18:14:15)
    7. - If it is Proton port, sends to Proton (bh526r, 18:14:35)
    8. JinLi join (jinli, 18:14:40)
    9. - Otherwise, goes to normal ML2 driver (bh526r, 18:14:53)
    10. 1.3 Dummy network, subnets etc. object will be used by Gluon ML2 driver to satisfies Nova-Neutron relationship (bh526r, 18:15:25)
    11. 1.4 Backend SDN controllers will create network, subnets etc. in our desired way (bh526r, 18:15:49)
    12. 1.5 South Bound stays the same etcd to communicate via Shim Layers with SDN Controllers, i.e.: (bh526r, 18:16:15)
    13. - Gluon ML2 core plugin will connect with etcd to determine port ownership (bh526r, 18:16:37)
    14. - Contrail will use existing plugin/mechanism driver, which is at the same level as etcd (bh526r, 18:16:55)
    15. - Contrail’s mechanism driver needs to work with Gluon ML2 Core Plugin (bh526r, 18:17:15)
    16. 1.6 Tom and Jin will look into it if this really works or not (bh526r, 18:17:41)
    17. 2. Group discussed how to explore Gluon ML2 Core Plugin (bh526r, 18:18:13)
    18. 2.1 Experiment with wrapping / sub-classing method (bh526r, 18:18:50)
    19. - If successful, done (bh526r, 18:21:19)
    20. Kamal Hussain (kamal___, 18:21:37)
    21. 2.2 If not successful, we try to fix ML2 by working with Neutron ML2 team (bh526r, 18:25:17)
    22. - If successful, done (bh526r, 18:25:37)
    23. 2.3 If not successful, we will use Original Gluon Model, i.e. Nova-Gluon-Neutron (bh526r, 18:25:58)
    24. - Customization for commercial deployment (bh526r, 18:26:19)
    25. Tom and Jin look into it if this really works or not (bh526r, 18:26:38)
    26. 3. Group discussed Demo Scope for OpenStack Summit in Barcelona, and agreed: (bh526r, 18:27:10)
    27. 3.1 Multiple SDN Controller Backend Interoperability – use Proton (bh526r, 18:27:35)
    28. - New interaction Model with Neutron, compared to demo in OPNFV Summit (bh526r, 18:27:58)
    29. - Create VM with two vNICs. Neutron Port (e.g Linux Bridge) and Gluon Port (OVS, vRouter, etc) (bh526r, 18:28:50)
    30. 3.2 New service on-demand (new Proton/APIs generated on-demand) (bh526r, 18:29:17)
    31. - Dummy service implementation (bh526r, 18:29:39)
    32. 4. Group also discussed that we need to clean up Nova Plugin so that we can have Gluon as Neutron Proxy, which is our model today (bh526r, 18:30:15)
    33. Group recognized that it would take long time. (bh526r, 18:30:38)
    34. Ian will work on it, and in his to-do list (bh526r, 18:30:52)

  4. New Blueprint Discussion (bh526r, 18:32:52)
    1. Based on F2F discussion, 2 Blueprints were created: (bh526r, 18:35:05)
    2. One is for Gluon ML2 Plugin BP: https://blueprints.launchpad.net/python-gluon/+spec/gluon-ml2-plugin. This is assigned to Tom (bh526r, 18:35:38)
    3. The other one is for Gluon Nova Plugin BP: https://blueprints.launchpad.net/python-gluon/+spec/gluon-nova-plugin. This is assigned to Ian (bh526r, 18:36:02)
    4. Gluon ML2 Plugin BP: https://blueprints.launchpad.net/python-gluon/+spec/gluon-ml2-plugin is approved unanimously (bh526r, 18:43:20)
    5. Bin will communicate with Ian to add more details in Nova Plugin BP (bh526r, 18:43:40)

  5. Tracking Progress of PoC (bh526r, 18:44:04)
    1. https://jira.opnfv.org/browse/NETREADY-21 (bh526r, 18:45:44)
    2. A JIRA ticket was created as PM tool for PoC: https://jira.opnfv.org/browse/NETREADY-21 (bh526r, 18:53:08)
    3. Note that it is only for PoC related tasks, including logistics and other non-coding tasks (bh526r, 18:53:35)
    4. Coding the documentation related tasks are referenced back to BP so that those will be tracked in Gerrit and BP/Launchpad in OpenStack (bh526r, 18:54:16)
    5. For example, a story of Gluon ML2 Driver in JIRA references back to Gluon ML2 Driver BP: https://jira.opnfv.org/browse/NETREADY-22 (bh526r, 18:55:21)
    6. Details of the Gluon ML2 Driver work will be tracked in Gerrit and BP/Launchpad (bh526r, 18:55:53)
    7. Once the coding and documentation are completed, JIRA will merely change ths status to "Resolved". (bh526r, 18:56:28)
    8. Meeting adjourned (bh526r, 18:58:14)


Meeting ended at 18:58:36 UTC (full logs).

Action items

  1. (none)


People present (lines said)

  1. bh526r (115)
  2. georgk (23)
  3. pcarver (12)
  4. KamilRenczewski (8)
  5. openstack (3)
  6. kamal___ (2)
  7. jinli (1)


Generated by MeetBot 0.1.4.