#opendaylight-meeting: MD-SAL interest call

Meeting started by colindixon at 16:00:10 UTC (full logs).

Meeting summary

  1. agenda bashing (colindixon, 16:00:14)
    1. https://meetings.opendaylight.org/opendaylight-meeting/2015/md_sal_ineterest_call/opendaylight-meeting-md_sal_ineterest_call.2015-07-14-16.05.html last week’s minutes (colindixon, 16:00:32)
    2. https://lists.opendaylight.org/pipermail/controller-dev/2015-July/009902.html the e-mail thead after last weeks meeting on the proposed new data store APIs (colindixon, 16:02:31)
    3. http://gist.asciidoctor.org/?c5f1027c657cda428a5d the proposed new data store APIs themselves (colindixon, 16:02:48)
    4. from last week, we had planned to talk about what we would do with the developer design forum time (colindixon, 16:03:08)
    5. https://wiki.opendaylight.org/view/Events:Be_Dev_Forum potential topics for the developer deisgn forum are here (colindixon, 16:03:32)
    6. TomP says they covered a bit about the new DataTree APIs at the end of the last meeting and agreed to talk about it more in person at the summit next week (colindixon, 16:06:26)
    7. https://wiki.opendaylight.org/view/Events:Be_Dev_Forum#Clustering_enhancements this will be discussed under this topic for those who are interested (colindixon, 16:07:09)

  2. how can we best use our time at the summit? (colindixon, 16:07:48)
    1. ttkacik shows the topics from the above link (colindixon, 16:08:45)
    2. ttkacik and moiz say that there is likely one long topic (clustering enhancement) listed above (colindixon, 16:09:21)
    3. ttkacik says there are also protocol-specific clustering topics, e.g., NETCONF and OpenFlow clustering (colindixon, 16:09:46)
    4. ttkacik another topic on upgrading the MD-SAL and migration from one version to another version (supporting multiple YANG model revisions) (colindixon, 16:10:23)
    5. colindixon notes that we don’t support multiple YANG model revisions via RESTCONF (tbachman, 16:10:27)
    6. colindixon asks whether transactionality in NETCONF is also a topic of interest (tbachman, 16:11:32)
    7. ttkacik says it’s of interest, but NETCONF already does 2-phase commits, and has done so since Hydrogen (if device supports it) (tbachman, 16:12:01)
    8. cross-data-center clustering may also be interesting to people (colindixon, 16:12:04)
    9. colindixon asks if we want to write code during the dev summit or are we just white-boarding? (tbachman, 16:12:28)
    10. ttkacik says we’re probably more interested in discussing cross-project concerns (tbachman, 16:12:50)
    11. moizer says we should add a topic for application clustering requirements discussion (tbachman, 16:13:28)
    12. ttkacik says application level clustering is an important topic and recommends it being its own topic (tbachman, 16:14:37)
    13. moizer will lead the application level clustering topic session (tbachman, 16:15:28)
    14. pantelis asks how long each session is (tbachman, 16:16:53)
    15. ttkacik says phrobb was going to look at the possible topics and attendees, and to try and schedule the sessions so that similar subjects don’t overlap — this might affect how long the sessions are scheduled for (tbachman, 16:17:38)
    16. colindixon asks phrobb how long the topics are going to be — should we be estimating lengths? (tbachman, 16:18:56)
    17. phrobb recommends estimating a length — also a relative number (e.g. base clustering session before application level clustering) (tbachman, 16:19:27)
    18. pantelis asks if we should combine openflow and NETCONF (tbachman, 16:20:21)
    19. ttkacik says he’d like to keep the openflow and HA clustering separate — is a big topic (tbachman, 16:20:34)
    20. moizer says we’ll first be discussing the use cases, and once we’ve done that, then we can talk about clustering enhancements, then finally applications (tbachman, 16:21:49)
    21. https://wiki.opendaylight.org/view/OpenDaylight_Controller:MD-SAL:Usage_Pattern_Questionare wiki page for clustering usage questionaire from ttkacik (tbachman, 16:22:31)
    22. phrobb says we have 12 hours for the sessions (tbachman, 16:33:35)
    23. colindixon would like to see a log normalization session — how to make log messages more uniform across projects (tbachman, 16:38:28)
    24. https://meetings.opendaylight.org/opendaylight-meeting/2015/tws/opendaylight-meeting-tws.2015-07-13-17.01.html (colindixon, 16:41:49)
    25. https://meetings.opendaylight.org/opendaylight-meeting/2015/tws/opendaylight-meeting-tws.2015-07-13-17.01.html at last week’s TWS call the SDN App SDK project presented a much more cut down version with just log aggregation analysis (colindixon, 16:43:03)
    26. phrobb is going to put things into an agenda format now — last call for topics! (tbachman, 16:43:35)
    27. moizer asks what will come out of these topics? bugs/enhancements? (colindixon, 16:44:33)
    28. ttkacik says modifications to release plans would be ideal (colindixon, 16:44:52)
    29. for controller/netconf/yangtools/mdsal would likely be enhancement bugs (colindixon, 16:45:06)
    30. for other projects, who knows (colindixon, 16:45:13)
    31. colindixon says that design documents also seem to make sense (colindixon, 16:45:22)
    32. moizer says that the specific work items are going to be bugs — so by the end, we should have the bugs on the release plan updated (tbachman, 16:45:35)

  3. inventory vs. topology models (colindixon, 16:49:21)
    1. as long as we have backward compatibility for all apps in Beryllium and deprecate the inventory model in Beryllium, we’re happy (colindixon, 16:49:46)
    2. colindixon says that you can eliminate so much code if you allow flow programming via topology instead of having to use both inventory and topology (colindixon, 16:51:47)
    3. colindixon and moizer both say that we should deprecate inventory in Be and remove it in B (colindixon, 16:52:09)
    4. ACTION: ttkacik and colindixon to talk to Keith about scheduling time on the TWS (colindixon, 16:52:43)
    5. TomP asks if we can deprecate a YANG model, ttkacik says yes, but he’s not sure that we pass it to the binding APIs, but if not we’ll add it (colindixon, 16:53:10)
    6. RESTCONF has no real way to signal deprecation (colindixon, 16:53:26)

  4. release plans (colindixon, 16:55:54)
    1. https://wiki.opendaylight.org/view/Simultaneous_Release:Beryllium_Release_Plan#Offset_0_Projects release plans for offset 0 projects are listed as drafts here (colindixon, 16:56:20)
    2. thanks to ttkacik for posting those (colindixon, 16:56:27)


Meeting ended at 20:06:53 UTC (full logs).

Action items

  1. ttkacik and colindixon to talk to Keith about scheduling time on the TWS


Action items, by person

  1. colindixon
    1. ttkacik and colindixon to talk to Keith about scheduling time on the TWS
  2. ttkacik
    1. ttkacik and colindixon to talk to Keith about scheduling time on the TWS


People present (lines said)

  1. colindixon (37)
  2. tbachman (34)
  3. odl_meetbot (5)
  4. ttkacik (0)


Generated by MeetBot 0.1.4.