#opendaylight-meeting: kernel projects

Meeting summary

1. agenda bashing (rgoulding, 16:04:32)
   1. yangtools 2.0.7 (rgoulding, 16:04:36)
   2. clustering csit tests (rgoulding, 16:05:11)
   
   
2. clustering csit tests (rgoulding, 16:06:04)
   1. 3-node controller test (rgoulding, 16:06:09)
   2. one bug that Vishal, Tom, and Jamo are working on (rgoulding, 16:06:56)
   3. https://jira.opendaylight.org/browse/CONTROLLER-1838 (rgoulding, 16:07:20)
   4. RESTCONF is lost for 20-30 seconds in this test (rgoulding, 16:07:38)
   5. is this expected? (rgoulding, 16:07:47)
   6. HTTP calls to RESTCONF time out (rgoulding, 16:08:19)
   7. lasts for 30+s (rgoulding, 16:08:25)
   8. Jamo claims he thinks they still hit the issue even with odl-restconf-noauth (rgoulding, 16:09:02)
   9. the script to adds cars is running with 10 threads (rgoulding, 16:09:45)
   10. possible queueing issue pointed out (rgoulding, 16:09:54)
   11. https://jenkins.opendaylight.org/releng/user/jluhrsen/my-views/view/controller%203node/job/controller-csit-3node-clustering-all-oxygen/ (rgoulding, 16:10:15)
   12. want to make ^^ stable (rgoulding, 16:10:20)
   13. all 10 failures in the above test are consistent and have happened for a month (rgoulding, 16:10:31)
   14. this appears to be the most comprehensive test for the clustering in the controller (rgoulding, 16:10:59)
   15. Victor brings up the point that we need to do an analysis of test cases and make sure the cases are valid and cover all cases (rgoulding, 16:11:32)
   16. do we need to add more test cases to cover other scenarios (rgoulding, 16:11:43)
   17. shague asks about what logs are useful for resolving issues with clustering (rgoulding, 16:12:48)
   18. rovarga thinks it is dependent on the issue (rgoulding, 16:12:55)
   19. shague asks are there any akka logs to keep running and collecting in background (rgoulding, 16:14:22)
   20. rovarga says not really (rgoulding, 16:14:27)
   21. right now trying to reproduce this with “tell-based” protocol, so AskTimeoutException isn’t the issue we are talkigna bout right now (rgoulding, 16:15:17)
   22. with the test in CONTROLLER-1838, they are enabling tell-based to see if it improves the situation. (rgoulding, 16:17:19)
   23. most of this landed in Carbon-SR1 timeframe (rgoulding, 16:19:00)
   24. hasn’t been used much since because someone needs to completely test it (rgoulding, 16:19:08)
   25. they will continue to test tell-based (rgoulding, 16:24:24)
   26. use case for writing to config/oper at once time-> the use case does not require writing to config/oper in the same transaction (rgoulding, 16:25:45)
   
   
3. yangtools 2.0.7 (rgoulding, 16:27:16)
   1. now that we are ready to bump to odlparent 3.1.2, we probably need to bump to yangtools 2.0.7 sometime before the release (rgoulding, 16:27:54)
   2. the RC should be ready in 10 minutes according to rovarga (rgoulding, 16:28:06)
   3. skitt has pushed 3 patches to test the bump (rgoulding, 16:28:17)
   4. yangtools changes could flush out model issues (rgoulding, 16:29:32)
   5. in cases of duplicate choices / cases (rgoulding, 16:29:39)

Action items

1. (none)

People present (lines said)

1. rgoulding (38)
2. odl_meetbot (3)
3. shague (2)
4. faseelak (1)