#opendaylight-meeting: TWS

Meeting summary

1. ODL Autorelease tool and process presentation (tbachman, 17:04:28)
   1. Giovanni is taking the lead. (Madhu, 17:04:42)
   2. autorelease is a process that tries to streamline the release process (tbachman, 17:04:56)
   3. autorelease process is required for a better release process due to all the issues that we faced during Hydrogen release. (Madhu, 17:06:19)
   4. slides : https://wiki.opendaylight.org/images/8/84/OpenDaylight_AutoRelease_TWS_Presentation.ppt (Madhu, 17:07:23)
   5. one such issue is inter-project dependencies (tbachman, 17:07:23)
   6. edwarnicke asks who was there for the “Monday of Doom” for the Hydrogen release. (tbachman, 17:08:56)
   7. rovarga notes it was Monday and Tuesday (tbachman, 17:09:07)
   8. The current release process has too much manual processing (tbachman, 17:09:44)
   9. is it really 2 days of doom.. it went on forever and we are still feeling the repercussions ;) (Madhu, 17:09:45)
   10. "As God as my witness, Monday will never happen again!" ;) (edwarnicke, 17:10:59)
   11. there were two categories of issues: versions-maven-plugin didn’t reliably do what was needed, and the Maven “out-of-the-box’ release process is complex and requires access to external git repos (tbachman, 17:11:47)
   12. the Giovanni's slide (https://wiki.opendaylight.org/images/8/84/OpenDaylight_AutoRelease_TWS_Presentation.ppt) is verbose enough in explaining the release-plugin issues :) (Madhu, 17:13:49)
   13. ttps://wiki.opendaylight.org/images/8/84/OpenDaylight_AutoRelease_TWS_Presentation.ppt) <== slides for presentation (tbachman, 17:14:59)
   14. edwarnicke asks why “git short SHA” is used in the versioning? (tbachman, 17:18:33)
   15. GiovanniMeo says that this makes it easy to identify a particular artifact (tbachman, 17:19:59)
   16. all the projects that get put in the same run will have the same day, month, year, etc., but will have a unique SHA (tbachman, 17:21:05)
   17. We could have used literal forms of identification, but they run the risk of ambiguity (tbachman, 17:22:50)
   18. rovarga asks if the git short SHA can be captured within the artifcats (tbachman, 17:23:21)
   19. GiovanniMeo says that’s already available in the pom.xml (tbachman, 17:23:37)
   20. rovarga asks if this can be captured into the actual jar. (tbachman, 17:24:00)
   21. GiovanniMeo says this is possible, and we can look to Eclipse as a reference (tbachman, 17:24:18)
   22. In the manifest of the .jar, the eclipse project includes a SHA for each single file of source used to create this artificat. (tbachman, 17:24:49)
   23. they also sign with GPG each of these signatures (tbachman, 17:24:59)
   24. regXboi notes that we’re already doing a (tbachman, 17:31:18)
   25. regXboi notes that we’re already using a “-n” for stable branches (tbachman, 17:31:39)
   26. discussion on human-readable vs. SHA defined identifiers (tbachman, 17:34:45)
   27. humans find it easier to have the -n conventions, but SHAs are the most specific for what an artificat is (tbachman, 17:35:11)
   28. Madhu points out that weekly releases aren’t necesarrily stable (tbachman, 17:38:20)
   29. This brings in the weekly release process in place & does not guarantee Stability of the artificats. (Madhu, 17:44:32)
   30. regXboi requests for making it obvious by adding WEEKLY in the artifact name (Madhu, 17:45:07)
   31. correction "-WEEKLY-" :) (Madhu, 17:45:47)
   32. GiovanniMeo will complete the rest of his presentation in Tomorrow’s MD-SAL call (tbachman, 17:48:12)
   33. Raghu clarifies the point that the Surekha's parent pom work and Giovanni's automated release process are complementary. Both these work are important & the version cleanup is needed. (Madhu, 17:50:14)
   
   
2. Automated Robot Integration test to verify FlowMod installation (Madhu, 17:50:27)
   1. https://wiki.opendaylight.org/view/CrossProject:Integration_Group starting link for Integration team (tbachman, 17:51:35)
   2. https://wiki.opendaylight.org/view/CrossProject:Integration_Group:Download_and_Run_System_Test (regXboi, 17:52:38)
   3. first is to clone the git repo (tbachman, 17:52:54)
   4. There are now two levels in their test suties folder (tbachman, 17:53:19)
   5. first level is main suite for a given distro (base, VTN, etc.) (tbachman, 17:53:35)
   6. second level is for the sub-suties that run within the main suite (tbachman, 17:53:46)
   7. Madhu asks whether the first level should be something like “virtualization”, and the second level should be something lk “VTN” (tbachman, 17:56:03)
   8. Testing at the first level means that a controller instance is launced with a given configuration (tbachman, 17:57:25)
   9. Tesitng at the second level can be done with an already-launched controller (tbachman, 17:57:39)
   10. Luis notes that they haven’t yet decided what this will look like for the Helium release. (tbachman, 17:58:03)
   11. the “ha” top level test dir isn’t yet ready (requires some LF infra support) (tbachman, 18:02:20)
   12. test group is working to get all of the test cases documented in testopia (tbachman, 18:05:31)
   13. every time a robot test is run, 3 reports are generated. Reports are easily viewed from a web browser (tbachman, 18:09:39)
   14. system tests using robot framework all run outside the controller, using REST (tbachman, 18:10:03)
   15. the test framework is currently in use in Jenkins (tbachman, 18:11:03)
   16. https://wiki.opendaylight.org/view/CrossProject:Integration_Group:FlowMod_Test (CASP3R, 18:11:43)
   17. https://lists.opendaylight.org/pipermail/integration-dev/2014-May/001148.html Integration hackfest (maberg, 18:15:44)
   18. team is currently testing with OVS, but is working with LF infra folks to add CPqD (tbachman, 18:19:15)

Action items

1. (none)

People present (lines said)

1. tbachman (50)
2. Madhu (20)
3. regXboi (8)
4. edwarnicke (7)
5. odl_meetbot (4)
6. maberg (2)
7. cdub (2)
8. CASP3R (1)
9. phrobb (0)