#opnfv-copper: Copper Weekly Meeting

Meeting started by bryan_att at 14:56:07 UTC (full logs).

Meeting summary

  1. Roll Call (bryan_att, 14:56:20)
    1. Bryan Sullivan (bryan_att, 14:56:30)
    2. Howard Huang (zhipeng, 14:59:46)

  2. Copper documentation (bryan_att, 15:04:00)
    1. I've been working to convert the docs to RST (bryan_att, 15:04:24)
    2. Once they are in the git repository i will send out a note for additional contributions (bryan_att, 15:04:59)
    3. in the short term you can see the document developing at my site http://bkaj.net/opnfv/copper/html/ (bryan_att, 15:05:27)
    4. this is using the sphinx RST renderer as is being used in the doctor project (bryan_att, 15:05:54)
    5. I've broken out the docs into use cases, architecture, and requirements sections (bryan_att, 15:06:33)
    6. the goal is to simplify contribution and get input from people already using gerrit etc (bryan_att, 15:07:32)
    7. for now I can be called upon as a reviewer and I will respond asap - less than 24 hours (bryan_att, 15:08:08)
    8. I'm also less concerned about minor editorial things and ensuring that group consensus is derived before commit merges (bryan_att, 15:08:47)
    9. I think the most important thing is that we develop the docs and review them periodically to ensure that they are going in the direction members want (bryan_att, 15:09:20)
    10. during the development phase there needs to be as low a barrier to contribution as possible - if anyone contributes something that someone else objects to, then an issue can be raised to address it (bryan_att, 15:10:12)
    11. that's most of the documentation update stuff - I should have the docs in git and the jenkins build setup by the end of this week (bryan_att, 15:11:18)

  3. OpenStack summit followup (bryan_att, 15:12:17)
    1. Next week I will be at ONS, and a subject will be the NB abstraction of configuration intent (bryan_att, 15:12:43)
    2. there's supposed to be a work item kicking off in ONF related to that, to define a grammar for intent (bryan_att, 15:13:16)
    3. I want us to begin to crystalize key use cases into a descriptive list of intent items, that we can then match to the existing NB interfaces and data formats (Yang etc) of VIMs (bryan_att, 15:14:26)
    4. That's one track of the followup - how do we express intent across VIMs (bryan_att, 15:14:52)
    5. A second track is how well the use cases match up to the capabilities of Congress and ODL projects (bryan_att, 15:15:18)
    6. analysis work on ODL projects has not yet started (bryan_att, 15:15:45)
    7. new intent project Keystone proposed in ONF (zhipeng, 15:16:12)
    8. Due to the need to get Arno up and running for use case testing of Congress first (bryan_att, 15:16:14)
    9. should it also be a Copper upstream ? (zhipeng, 15:16:25)
    10. That's my #1 goal - a real testable platform to validate use cases (bryan_att, 15:16:42)
    11. But I encourage any assessment of how configuation intent is propagated to ODL / Contrail / ONOS etc through OpenStack as the orchestrator (bryan_att, 15:17:33)
    12. I have an adequate handle on Congress, but none on the ODL support (bryan_att, 15:17:57)
    13. I'll be adding spaces to the docs so that this analysis can be documented as people progress it (bryan_att, 15:18:34)
    14. The 3rd takeaway is what closed-loop systems can help us implement reactive policy (bryan_att, 15:19:29)
    15. I want to align this in particular with Doctor, Promise etc - there is a lot in common, e.g. approaches (bryan_att, 15:20:27)
    16. such as listening to message busses, or subscribing to specific events at publlishers (bryan_att, 15:21:02)
    17. That leads to possible considerations for refactoring Copper and other projects so that the common aspects are merged in some place/project. Any ideas how we can go about that? (bryan_att, 15:22:09)
    18. The intent abstraction goal is also a common aspect, like closed-loop support (bryan_att, 15:22:52)
    19. Closed-loop means there is an event at some place, and a listener at another place gets and event, and takes some action in response (bryan_att, 15:24:52)
    20. As compared to just handling events at the source (bryan_att, 15:25:13)
    21. if polices are applied by the entity that established them, e.g. a VIM, which also locally discovered the event that affects policy, that is local enforcement as compared to closed-loop enforceent (bryan_att, 15:26:33)
    22. That's about all I need to say for now re the takeaways - there are three main takeaways, two of which are likely in common with the other projects (intent expression, and closed-loop methods) (bryan_att, 15:27:16)
    23. the need for a testbed is copper-specific (e.g. layering on of Congress through Ansible) (bryan_att, 15:27:44)
    24. and the ability to locally detect configuration policy violations is also Copper-specific (bryan_att, 15:28:35)
    25. So in summary for release 2 ... (bryan_att, 15:29:00)

  4. Release 2 workplan (bryan_att, 15:29:16)
    1. What i want to do is start a cross-project dialog on factoring out common aspects of policy-driven projects, so we are investigating these things is disconnected projects (bryan_att, 15:30:04)
    2. or (are not) (bryan_att, 15:30:28)
    3. The other main goal now that we have Arno is to develop and validate use cases, in a testbed (bryan_att, 15:31:18)
    4. That will involve assessment of ODL projects and other OpenStack projects related to config policy (bryan_att, 15:32:02)
    5. With two main focuses (use case testing and local config enforcement) I think we can make good progress before the liberty deadline and ODL release deadline (bryan_att, 15:32:58)
    6. so in summary I'm proposing that OPNFV work to collect common project aspects into a new project or centralize them somehow, while ensuring that use cases (e.g. config, fault mgmt, scaling, reservation) are equally addressed (bryan_att, 15:34:26)
    7. or other policy related rojects just output the requirements to Copper (zhipeng, 15:35:07)
    8. While in Copper we focus back on the use case analysis and local policy enforcement (as a start, expanding to closed-loop once we have assessed how to do this in a common way) (bryan_att, 15:35:42)
    9. I don't propose that Copper be the common place for aspects in common, but that is certainly an ioption. (bryan_att, 15:36:17)
    10. What I want is for the community to determine where is the best place to address these aspects, e.g. where is the most momentum (bryan_att, 15:36:49)
    11. If we can use this as a means to amp up participation in Copper then fine (bryan_att, 15:37:09)
    12. But I assume that it's open to discussion (bryan_att, 15:37:21)
    13. to ONS preso on OPNFV (iben, 15:41:07)
    14. https://wiki.opnfv.org/start#events (iben, 15:41:08)
    15. - https://wiki.opnfv.org/_media/ons-opnfv-spirent-preso-ir150609.pptx (iben, 15:42:02)


Meeting ended at 15:42:32 UTC (full logs).

Action items

  1. (none)


People present (lines said)

  1. bryan_att (70)
  2. iben (24)
  3. zhipeng (11)
  4. collabot (3)


Generated by MeetBot 0.1.4.