#opnfv-meeting: OPNFV Q1 Hackfest

Meeting summary

1. 1. rprakash (rprakash, 16:05:13)
   2. David McBrick comes from Qualcom (rprakash, 16:09:51)
   3. agenda https://etherpad.opnfv.org/p/Q1'2016_Hackfest (rprakash, 16:11:02)
   4. announcing release of Brhmaputra (rprakash, 16:11:50)
   5. Heather describes the Brahmaputra release infographic showing how the community grew (ildikov_, 16:13:16)
   6. https://wiki.opnfv.org/releases/brahmaputra (rprakash, 16:14:01)
   7. Tomi from Nokia gets as best Coder Developer for OPNFV and is part of Doctor (rprakash, 16:14:36)
   8. for Code Developement Tomi Juvonen (rpaik, 16:14:48)
   9. Documentation Award also goes to Ryota Mibu (rpaik, 16:16:47)
   10. Dave Neary , Smith/Mark, Biert/Peter, Lee/Royta Mibu for Collabortaion awards (rprakash, 16:16:56)
   11. Integration award Cathy and Narinder Gupta (rprakash, 16:17:44)
   12. Future technical community events discussion (rpaik, 16:18:09)
   13. OPNFV summit on June 20-21at Berlin. Germany (rprakash, 16:18:48)
   14. the question is should we continue to co-locate with other events? (rpaik, 16:19:16)
   15. OPtions for H2 2016 ODL Seattle 27-29 Seattle, Plugfest in France (TBD) , LF Summit 20-24 Toronto Canada...choose one (rprakash, 16:20:55)
   16. IETF 97 at Seoul Nov 13-18 in Korea.. another option (rprakash, 16:23:00)
   17. comment that co-location with plugfest maybe a good idea (rpaik, 16:23:18)
   18. plugfest 2 invite from Orange in EU need discussions plus one more at IETF 97 in Korea (rprakash, 16:25:17)
   19. suggestion to start a poll on 2H events with the technical community (rpaik, 16:26:32)
   20. Bryan Sullivan wanted to know who are particpating where to maximize particpation and prefers ODL (rprakash, 16:26:37)
   21. Chris too likes that (rprakash, 16:26:53)
   22. OpenStack Summit-Barcelona is in the last week of October (rpaik, 16:28:34)
   23. suggestion to have pre-OpenStack discussion regarding Blueprint planning within the community (rpaik, 16:29:43)
   24. another option is Openstack In Barcelone, Spain - October 24-28, 2016 (rprakash, 16:29:45)
   25. Release C is Colorado (rprakash, 16:31:41)
   26. suggestion to schedule the afternoon breakouts around Colorado release planning discussion (rpaik, 16:32:13)
   27. changes : 4:30 - 6:00: Release C planning (rprakash, 16:34:50)
   
   
2. Brahmaputra retrospectives (rpaik, 16:35:09)
   1. Chris is looking for Constructive complaints for Release C plan for what we laerned in B (rprakash, 16:36:39)
   2. https://etherpad.opnfv.org/p/c-release-brainstorming Colorado release plan Etherpad (rpaik, 16:36:52)
   3. David McBrick to ask questions on feedback from Release B (rprakash, 16:39:14)
   4. q1. WHoa was inolved in testing Brahmaputra? (rprakash, 16:39:39)
   5. what was comprehensive test coverage (rprakash, 16:40:07)
   6. trying to do everything all at once (esp. for testing) was a major challenge (rpaik, 16:40:41)
   7. Tim redhat say too many parallel tasks hence we barely finished creating tests and hence covergae was minimum and problem was lot of baremetals and too much at last movement (rprakash, 16:41:33)
   8. test prep activies should be done earlier in the prep cycle (rpaik, 16:42:02)
   9. David says to figure out what test cases before rather than in late in release cycle (rprakash, 16:42:14)
   10. this improved from A to B and needs to improve more in Colorado (Alan_, 16:42:57)
   11. Dependency between individual project and testing was evdient in the Projects as everyone was tied to Func_test and hence Unit testing should be added (rprakash, 16:44:20)
   12. interdepencies between various projects requires more unit testing (e.g. of feature projects) earlier (rpaik, 16:44:29)
   13. Installer by Installer was painful process so need some methodology (rprakash, 16:44:41)
   14. everybody seems to do differently (rprakash, 16:44:58)
   15. David say we need consistancy, but other say differentaite between Unit tests from Production environment (rprakash, 16:45:51)
   16. chris - normalize CI pipeline so same methodology from unit to system (Alan_, 16:46:17)
   17. Chris say we need to streamline between CI and testing (rprakash, 16:46:23)
   18. frankbrockners adds that scenarios were introduced late in the cycle (rpaik, 16:47:05)
   19. Frank says concept of Scenerio needed to be completed with feature before taking up to testing (rprakash, 16:47:11)
   20. Frank says integration was in a rush (rprakash, 16:47:29)
   21. Frank not everything works at same pace with Multiple Scenerios (rprakash, 16:47:54)
   22. Two weeks of Testing and Two weeks of Integration is not sufficient (rprakash, 16:48:26)
   23. Frank - define Colorado scenarios early and each works at own speed (Alan_, 16:48:41)
   24. Frank says so we should not club synchronizing Scenerios (rprakash, 16:48:56)
   25. frankbrockners add that we should schedule scenarios independent from each other (rpaik, 16:49:19)
   26. - Chris: Unit and System and isolated (and unit comes from upstream) but system has dependencies on unit (Alan_, 16:52:21)
   27. rprakash said test at MS3 would be helpfull (rprakash, 16:52:53)
   28. need to have clearer milestones for testing (rpaik, 16:52:57)
   29. Chris points that lot of testing is in upstream and hence Yardstick needs differentaite for OPNFV Scenerios (rprakash, 16:53:40)
   30. ChrisPriceAB asks if we can find a way to combine scenarios (rpaik, 16:54:16)
   31. Chris susggests reducing Scenerios (rprakash, 16:54:19)
   32. Tim says tie Configuration to Scenerios instead of Calling Confguration of Doctor in multiple Scenerios (rprakash, 16:55:01)
   33. - Benefit of this is that features are isolated to disjoint scenarios currently (Alan_, 16:55:03)
   34. dradez asks if we can have scenarios based on configurations vs. features (rpaik, 16:55:36)
   35. we need to be able to deploy any scenario on any POD (rpaik, 16:56:48)
   36. arturt notes that we don’t have enough Pharos hw resources that we can easily consume (rpaik, 16:58:17)
   37. - Chris - we were under-resourced (<20%) on both community and LF resources - that is, we could not get them "on demand" or soon enough (Alan_, 16:59:31)
   38. infratructure support for Pharos resources is an area for improvement (rpaik, 16:59:55)
   39. need more lab support besides Aric from LF (rprakash, 17:00:18)
   40. Tim says Development testing needs early look -means unit testing asks David (rprakash, 17:01:49)
   41. question is raised if we can complete development test prior to production tests? (rpaik, 17:02:01)
   42. Trevor says its Fetaure testing (rprakash, 17:02:04)
   43. Doctor project rolled their testing in Functest and that worked well (rpaik, 17:05:09)
   44. Saves time if we get help from platform folks as how use case can be tested (rprakash, 17:06:26)
   45. end user getting involed is a good (rprakash, 17:06:45)
   46. trozet notes that for Apex they need to add test cases to make sure that a new feature works (vs. just deploying) (rpaik, 17:08:19)
   47. Tim says Apex makes sure that add func_test to esnure part of system like Horizon end point access (rprakash, 17:08:24)
   48. installers need to add some of the “basic” functest suites (rpaik, 17:10:53)
   49. Installers have common Tempest tests and can be used as minimum smoke test plus validate more seperately to save test run time (rprakash, 17:11:37)
   50. Scenerio we need to link to be able to system level plus user level testing (rprakash, 17:14:02)
   51. frankbrockners adds that configuration file for functest is another improvement area (rpaik, 17:14:43)
   52. Frank says having proper configuraion files is better and says Morgan is looking at that for cleaning up (rprakash, 17:14:59)
   53. Chris says YardStick is targeted at system level whereas Func_Test targets some features (rprakash, 17:15:54)
   54. Smoke test must include some of the Func_test fro example (rprakash, 17:16:46)
   55. chris - maybe need to be giving specs to functest what it should provide (Alan_, 17:16:47)
   56. All - realization that scenario definitions did not have an owner - how should this work in C release? (Alan_, 17:17:40)
   57. discussion on where scenario configuration files should be (currently they’re with the installers) (rpaik, 17:18:28)
   58. we need to pull some from installer and put in some other scenerio to simplify the installation (rprakash, 17:18:36)
   59. Ian say jumping between Scenerios and Function should be minimum (rprakash, 17:20:17)
   60. Reason - features will be "smoke test passed and more" and scenario testing, while still hard, will be less frequent (Alan_, 17:22:31)
   61. ChrisPriceAB notes that there is a design guide for all testing projects (rpaik, 17:23:05)
   62. Producing an Artifact is being used by installer and so we should be able to merge some tests to staging to complete some system testing (rprakash, 17:24:29)
   63. Chris says we have documents for Yardstick and Func_Test for whitebox plus vping and others refer (rprakash, 17:26:33)
   64. https://wiki.opnfv.org/yardstick (rprakash, 17:26:37)
   65. https://wiki.opnfv.org/releases/c (rprakash, 17:27:31)
   66. we also need to discuss upstream planning for the D-release (e.g. OpenStack Newton) (rpaik, 17:27:36)
   67. are we missing anything in OpenStack (rprakash, 17:28:17)
   68. - Purpose - to not just be consuming everything but saying what bridges the gaps (Alan_, 17:28:46)
   69. consider Orchestartion not just Tacker and other upstream can do same besides OpenStack projects (rprakash, 17:34:43)
   70. Bryan says Tacker is for Test case for Generic VNF but commercial can be different (rprakash, 17:36:27)
   71. Bryan says have Package thatw works and add commercial stuff aove that (rprakash, 17:37:11)
   72. need to solicit input from the large community via mailing lists for feedback to upstream communites (rpaik, 17:37:33)
   73. Ian says we need to plan for upstream in OpenStack in next 4-6 weeks (rprakash, 17:38:04)
   74. ChrisPriceAB suggests working groups to converges some of the individual project discussions/activities (rpaik, 17:38:52)
   75. Chris talks up about NFVReadyNeSs and see how all this is going to look , instead having 40 projects can we get them to merge (rprakash, 17:38:58)
   76. workgroups proposal by Chris to get consensus between Projects is a good idea and collect and start common purpose represenation for planning (rprakash, 17:41:18)
   77. - Suggestion to make this process happen through WG's submitting proposals to the TSC (not forcing proj merger) (Alan_, 17:41:21)
   78. 20 minutes break now (rprakash, 17:41:45)
   79. rprakash (rprakash, 18:07:30)
   80. getting re-started in about a minute (rpaik, 18:07:59)
   
   
3. Development aand Testing Infrasructure (rprakash, 18:07:59)
   1. presented by Trevor Cooper & Matthew Li (rpaik, 18:08:24)
   2. Jenkins-ci what should be included in evrify job (rprakash, 18:10:11)
   3. unitest to ensure Project is in order (rprakash, 18:11:07)
   4. what is time for tests - 1 hour , 10 hour? (rprakash, 18:11:44)
   5. Not hours and Days but few minutes (rprakash, 18:12:08)
   6. Matthew notes that goal should be minutes (rpaik, 18:12:15)
   7. improvement 2 is to test in local env before adding patch to Project to ensure minimize reun times for progress in CI (rprakash, 18:14:16)
   8. submit changes after testing in local environment (rprakash, 18:14:41)
   9. another option is to integrate with community installers and run the test cases once patch is set with related code (rprakash, 18:16:05)
   10. community environmenmts like Intel, Ericsson, Huawei .... for installers in differet PODs (rprakash, 18:17:19)
   11. Improvement 3 is adding Dashboard (rprakash, 18:18:55)
   12. there are individual testing dashboards in addition to the test board that went live on opnfv.org (rpaik, 18:21:44)
   13. Frank says we should look at moving from Hand Crafted to automated dashboard (rprakash, 18:23:14)
   14. due to lack of test folks at Hackfest here we need to take it later in test workgroup or teams (rprakash, 18:24:34)
   15. Mathew also discusses Nodepool as a tool for lar resource management tool (rpaik, 18:25:33)
   16. Lab resource pool using NodeLable Parameter Plugin, but needs OpenSatck environment alternate being considered or Virtual environment and Fatih working on it with his Releng coleagues (rprakash, 18:27:37)
   17. Dev?Test Infra Goal/Vision (rprakash, 18:28:16)
   18. Lab Visibility: Capability, Usage, (rprakash, 18:28:44)
   19. Configuration Management (rprakash, 18:28:56)
   20. Common Inventory for Installer, POD Cobfiguration, Common Platform tests need development (rprakash, 18:30:12)
   21. Trevor notes that supporting a bare metal environement requires a lot of resources (rpaik, 18:32:17)
   22. and better support is required (rpaik, 18:32:30)
   23. Onboarding requires addressing (rprakash, 18:33:22)
   24. https://wiki.opnfv.org/pharos_laas (rprakash, 18:33:27)
   25. lab-as-a-service will need Infrastructure support - Lf Lab support, CI, Troubleshooting/sysadmin (rprakash, 18:35:08)
   26. aricg poses the question of providing potential users with direct access to jenkins toolchain (rpaik, 18:35:45)
   27. look at MaaS which was co-ordinating to ensure that Labs are utilized well and Iben and Arthur can lookback at that option (rprakash, 18:37:18)
   
   
4. OPNFV Plugfest (rpaik, 18:37:40)
   1. Aric has good question and may be we need to take this at Pharos meeting (rprakash, 18:38:21)
   2. Lincoln & Hongbo presenting the Plugfest update (rpaik, 18:39:56)
   3. Plugfest will take place during the week of May 9th at CableLabs in Louisville, CO (rpaik, 18:41:07)
   4. Lincoln itroduces Plugfest WG and Hongbo and Ray (rprakash, 18:41:25)
   5. http://events.linuxfoundation.org/events/opnfv-plugfest (rprakash, 18:41:30)
   6. Provide event/Opportunity for tesing OPNFV and measure results (rprakash, 18:42:16)
   7. Keyfoucs ...1. Deployment, 2. Network Integration and 3. VNF testing over Platform (rprakash, 18:42:52)
   8. plugfest is open to both members and non-members of OPNFV (rpaik, 18:43:08)
   9. an Open event and any one can particpate not just OPNFV members (rprakash, 18:43:16)
   10. critical path right now is test plan development (rpaik, 18:44:06)
   11. Plufest Timelines - Rules of Engagement (ROE) like NDA sorted ot, and planning tests for now, followed by exceution and some white paper (rprakash, 18:44:17)
   12. http://events.linuxfoundation.org/events/opnfv-plugfest/program (rprakash, 18:45:17)
   13. Hongbo presenting the Technical details (rprakash, 18:45:51)
   14. Basic Scenerio and Extended Scenerio (rprakash, 18:46:13)
   15. Basic hardware/software test for OPNFV Platform beyond Pharos Scenerios to cover interoperability (rprakash, 18:47:27)
   16. Flow -> Pharsos ->OPNFV installler ->B-Release -> Test Projects-> VNF and Life Cycle (rprakash, 18:48:46)
   17. Vendors flow: Hardware Vendor -> Installer Vendor -> Platform Vendor -> Tesdt vendor -> VNF Vendor (rprakash, 18:49:52)
   18. Morning Session (Setup Only all Basic), Afternoon session (Test Secnerios and fianlly any exteded services testing) (rprakash, 18:50:50)
   19. test cases under consideraion OPFV testing, SDN testing, VNF testing, Orchestartion testing are current thinking (rprakash, 18:51:54)
   20. Nextstep Plugfest Test development requesting Particpation for Plugfest for Infrastrucure and overall communication (rprakash, 18:53:16)
   21. https://www.regonline.com/Register/Checkin.aspx?EventID=1815505 (rprakash, 18:54:51)
   22. Join Divetail meeting Friday 1400:1500 UTC (rprakash, 18:56:40)
   23. https://global.gotomeeting.com/join/458547813 (rprakash, 18:56:45)
   24. IRC channel: IRC channel: #opnfv-meeting@ Freenode (Web Chat) (rprakash, 18:57:09)
   25. Plugfest should not replicate what other people are doing as per Frank so we focus on OPNFV what it does and add more as we move to Release C and D etc (rprakash, 18:58:24)
   26. want to tie in what we do in Plugfest to OPNFV and not replicate what others have done (e.g. ETSI, NIA, etc) (rpaik, 18:58:28)
   27. we should see how we can bring in MANO testing in future lets work through future Pludfests (rprakash, 19:00:35)
   28. important to remember that this is our first plugfest and will be a big learning opportunity (rpaik, 19:01:08)
   29. This first one should be a boot from ground and see how we can accomodate and inclusive as much as possible (rprakash, 19:01:15)
   30. This is setting the baseline for Plugfest as Ray Nugent mentioned (rprakash, 19:02:25)
   31. based on registarion may limit the partcipation and Ray Pike mentions and have a blog to promote the plugfest (rprakash, 19:03:55)
   32. 20 membership companies sign that itself will be huge (rprakash, 19:04:48)
   33. now breaking for lunch (rpaik, 19:05:06)
   34. Breaking for Lunch (rprakash, 19:05:19)
   35. re-starting in about a minute (rpaik, 20:08:39)
   
   
5. Colorado release planning (rpaik, 20:08:53)
   1. https://wiki.opnfv.org/releases/colorado?s[]=colorado&s[]=release Colorado release page (rpaik, 20:11:26)
   2. rprakash (rprakash, 20:15:38)
   3. Tim notes that for his feature project, installers were moving targets for the Brahmaputra release (rpaik, 20:15:50)
   
   
6. OPNFV Release Planning (rprakash, 20:15:51)
   1. Chris starts discussions on Mile Stones and release dependancy (rprakash, 20:17:09)
   2. there is a suggestion to have feature freezes first (rpaik, 20:18:37)
   3. upstream dependency must be understood early (rprakash, 20:23:08)
   4. before Milestone 1, upstream dependencies must be understood (rpaik, 20:27:42)
   5. Genesis can be used as a vehicle to comprehend upstream dependencies (rpaik, 20:29:34)
   6. Frank to talk about OPNFV Release and Scenerio Release (rprakash, 20:31:07)
   7. OPNFV test components and features at system level (rpaik, 20:32:43)
   8. frankbrockners asks if there should be release schedules per scenarios (rpaik, 20:34:42)
   9. per scenarios you could have fearure complete, install ready, test ready, integration ready, and release ready (rpaik, 20:36:38)
   10. try to combine upstream patches in “one OPNFV version of upstream” whenever possible (rpaik, 20:51:45)
   11. Scenerio based rlease within Release C being proposed by Frank and Chris (rprakash, 20:54:39)
   12. some debate on experimental scenerio vs. Production based scenerios (rprakash, 20:55:08)
   13. this proposal for scenarios based release allows for quicker feedback to upstream developer communities (rpaik, 20:56:27)
   14. Chris wants to avoid labeling it like HA, NOHA, SDN, NOSDN per installer and with 4-5 intalers this makes it 20 scenerios for now (rprakash, 20:57:09)
   15. so we need to get a team working on scenerio based cycle within OPNFV to track all scnerios as compositions of components and their configuration (rprakash, 20:58:30)
   16. As we get to Scenerio maturity we may be able to do away with lables like we have now HA, NoHA etc etc. (rprakash, 21:00:19)
   17. idea to create a mechanism for projects to create an RPM for deployment (rpaik, 21:11:29)
   18. consensus on the scenario-based release proposal (rpaik, 21:13:57)
   19. Frank proposed Scenerio based relase and most agreed to it and Chris said despite that we still need to manage OPNFV release based in parallel like Release B (rprakash, 21:14:36)
   20. Sometimes in July we may have a stable release plan and in that few scenerios that my qualify and moeet the goals of OPNFV C Release (rprakash, 21:15:51)
   21. Get Scnerio tested early to get requirements upstream, but at the end of the day relase freez at a given time must include the scenerios that make it to date of freeze and hence release (rprakash, 21:39:28)
   22. Ofee break before we gather for Project Breakout sessions (rprakash, 21:49:24)

Action items

1. (none)

People present (lines said)

1. rprakash (141)
2. rpaik (78)
3. collabot` (14)
4. Alan_ (12)
5. ildikov_ (1)
6. bryan_att (1)
7. narindergupta (1)
8. braynn_att (0)
9. ChrisPriceAB (0)