#opendaylight-meeting: kernel projects

Meeting started by rgoulding at 17:00:25 UTC (full logs).

Meeting summary

  1. agenda bashing (rgoulding, 17:00:29)
    1. https://git.opendaylight.org/gerrit/#/c/67283/ (rgoulding, 17:00:41)
    2. https://git.opendaylight.org/gerrit/#/q/topic:simple-dist_web (rgoulding, 17:00:56)
    3. Cluster isolation scenario (rgoulding, 17:01:11)
    4. blockers (rgoulding, 17:01:57)
    5. https://lists.opendaylight.org/pipermail/mdsal-dev/2018-March/001517.html (rgoulding, 17:03:01)

  2. Oxygen Blocker Review (rgoulding, 17:04:01)
    1. ODLPARENT-139: likely to be downgraded to major since there is a workaround (rgoulding, 17:05:24)
    2. NETCONF-520: reduced severity and will probably close (api change only) (rgoulding, 17:06:52)
    3. NETCONF-521: may be working but we need to test before we can reduce severity or close (rgoulding, 17:07:29)
    4. no other blockers for oxygen were identified by meeting participants (rgoulding, 17:08:36)

  3. Josh's Websocket Uri should have scheme wss if registration is https (rgoulding, 17:09:01)
    1. https://git.opendaylight.org/gerrit/#/c/67283/ (rgoulding, 17:09:12)
    2. there are build failures currently happening with the aforementioned patch (rgoulding, 17:10:01)
    3. some master netconf patches are successfully building, and it is unsure at this time whether the failures are related to this patch (rgoulding, 17:11:19)
    4. ACTION: vorburger to start a thread on mdsal-dev and netconf-dev related to broken master (rgoulding, 17:15:10)
    5. https://git.opendaylight.org/gerrit/#/q/topic:simple-dist_web (rgoulding, 17:15:34)
    6. package name will be left as is in the current patch set, with “aaa” in the package name (rgoulding, 17:18:42)
    7. the "mature" netconf project actually violates the separate namespace rule already (vorburger, 17:20:13)
    8. ACTION: netconf-aaa-plugin violates this aforementioned namespace rule, so rgoulding to file an issue (rgoulding, 17:20:20)
    9. tpantelis points out that restconf has the same sort of issue too (rgoulding, 17:20:40)
    10. rovarga believes it should be okay to leave this “test” dependency for the time being (rgoulding, 17:23:38)
    11. ACTION: rgoulding to investigate getting rid of embed-dependency of h2 (rgoulding, 17:28:53)
    12. ACTION: rgoulding to pull down Web API changes and then test RESTCONF to see if it works with these changes (rgoulding, 17:35:18)

  4. Binding V1 codegen in Fluorine (rgoulding, 17:37:20)
    1. https://lists.opendaylight.org/pipermail/mdsal-dev/2018-March/001517.html (rgoulding, 17:37:21)
    2. code generated by fluorine codecs would not load in the oxygen container, basically (rgoulding, 17:38:08)
    3. rovarga wanted to raise awareness since no one really said anything on the mailing list (rgoulding, 17:38:25)
    4. this one doesn’t change any of the API contracts so consumers will not need fix-up patches (rgoulding, 17:39:14)
    5. (just changes class internals) (rgoulding, 17:39:20)
    6. https://lists.opendaylight.org/pipermail/tsc/2018-March/009053.html (rovarga, 17:39:22)
    7. this one would require changes to existing users (rgoulding, 17:39:48)
    8. this is a change to md-sal apis (rgoulding, 17:40:33)
    9. “2) Replace org.opendaylight.yangtools.yang.binding.Identifiable.getKey()” —> this will break a lot of code, but it is a mechanical change (rgoulding, 17:41:24)
    10. this will break projects outside of our ODL ecosystem that rely on MD-SAL java apis (rgoulding, 17:42:21)
    11. none of these changes should be super intrusive, and will extend the life of the binding spec v1 (rgoulding, 17:43:59)
    12. the fallout is obvious and easy to fix (rgoulding, 17:44:17)
    13. there is a draft calendar for fluorine (rgoulding, 17:53:04)
    14. https://logs.opendaylight.org/releng/vex-yul-odl-jenkins-1/docs-rtd-verify-fluorine/27/html/release-process/managed-release.html (klou, 17:53:06)

  5. Cluster isolation scenario (rgoulding, 17:55:56)
    1. https://jira.opendaylight.org/browse/CONTROLLER-1817 (rgoulding, 17:56:32)
    2. quarantine should restart bundle 0 (rgoulding, 17:57:50)
    3. original testing was done by Gary Wu with a much older version of akka (rgoulding, 17:58:09)
    4. there were some issues regarding how to restart karaf container found (when we were supposed to restart bundle 0 we weren’t really) (rgoulding, 17:58:57)
    5. this wasn’t an issue in Nitrogen (rgoulding, 17:59:06)
    6. to clear akka state (restart JVM) you need to kill bin/karaf, then restart, which was not possible in karaf 3.0.X (rgoulding, 18:00:46)
    7. there was a property set to restart karaf, which did not actually clear everything out (rgoulding, 18:01:06)
    8. not sure why sometimes the event is sent only sometimes (rgoulding, 18:01:46)
    9. advice is to dig into what akka is doing and why it is not working with our implementation (rgoulding, 18:02:36)
    10. akka has a debug mode in the akka configuration to help create more verbosity in the log (rgoulding, 18:03:23)


Meeting ended at 18:06:14 UTC (full logs).

Action items

  1. vorburger to start a thread on mdsal-dev and netconf-dev related to broken master
  2. netconf-aaa-plugin violates this aforementioned namespace rule, so rgoulding to file an issue
  3. rgoulding to investigate getting rid of embed-dependency of h2
  4. rgoulding to pull down Web API changes and then test RESTCONF to see if it works with these changes


Action items, by person

  1. rgoulding
    1. netconf-aaa-plugin violates this aforementioned namespace rule, so rgoulding to file an issue
    2. rgoulding to investigate getting rid of embed-dependency of h2
    3. rgoulding to pull down Web API changes and then test RESTCONF to see if it works with these changes
  2. vorburger
    1. vorburger to start a thread on mdsal-dev and netconf-dev related to broken master


People present (lines said)

  1. rgoulding (50)
  2. odl_meetbot (3)
  3. klou (1)
  4. rovarga (1)
  5. vorburger (1)


Generated by MeetBot 0.1.4.