#opnfv-testperf: testperf

Meeting started by dfarrell07 at 13:58:27 UTC (full logs).

Meeting summary

    1. CPerf (dfarrell07) is chair, VSPerf (acm) is taking minutes (for first 30 mins or so) (dfarrell07, 13:58:55)
    2. https://wiki.opnfv.org/display/meetings/TestPerf Agenda in the normal place (dfarrell07, 13:59:04)
    3. Al Morton (acm, 14:00:13)
    4. Feel free to #info in while we wait on someone to start the gotomeeting (dfarrell07, 14:00:25)
    5. Morgan Richomme (morgan_orange, 14:00:33)

  2. Action follow-up (dfarrell07, 14:02:25)
    1. (last week #action) morgan_orange initiate a wiki page on POD allocation for test projects (short/mid/long term) after the C release (dfarrell07, 14:02:33)
    2. (last week #action) kubi001 morgan_orange plan an update on grafana/ELK in September (dfarrell07, 14:02:39)
    3. yuyang (yuyang, 14:02:40)
    4. (last week #action) mbeierl to summarize stable branch guidelines for testperf members (dfarrell07, 14:02:45)
    5. (last week #action) yuyang morgan_orange kubi001 create patch and put test people in review (dfarrell07, 14:02:50)
    6. (last week #action) jose_lausuch bryan_att sync how to push results in artifact from copper (dfarrell07, 14:02:55)
    7. kubi (kubi001, 14:02:57)
    8. no progress on POD allocation (acm, 14:03:22)
    9. ACTION: (carry-over, after release)kubi001 morgan_orange plan an update on grafana/ELK in September (dfarrell07, 14:03:26)
    10. POD #action will happen after release (dfarrell07, 14:04:10)
    11. Finish the release first (on POD alloca) (acm, 14:04:17)
    12. (done) jose_lausuch bryan_att sync how to push results in artifact from copper (dfarrell07, 14:04:25)
    13. docs review inter-project has started (dfarrell07, 14:04:47)

  3. Colorado status (dfarrell07, 14:05:12)
    1. Colorado Status (acm, 14:05:16)
    2. functest reports that things are fairly green for C status, were not in such good shape with A and B, so progress, some bugs to be fixed but that's expect (dfarrell07, 14:06:13)
    3. All installers and scenarios seem to be making progress (dfarrell07, 14:06:30)
    4. same pod for release and feature test - issues (acm, 14:07:11)
    5. Feature projects need time for integration, are causing some false-negatives errors that installers are seeing because feature projects are using same pods (dfarrell07, 14:07:18)
    6. red status in Jenkins is not end-of -world... (acm, 14:07:39)
    7. Better to get feedback about what's working and not, vs green on smaller set of things (dfarrell07, 14:07:52)
    8. kubi001 from yardstick gives updates (dfarrell07, 14:09:11)
    9. https://wiki.opnfv.org/display/yardstick/Yardstick+Colorado+CI+Status (kubi001, 14:09:16)
    10. Yardstick - new wiki to track stable branch (above) (acm, 14:09:24)
    11. https://wiki.opnfv.org/display/yardstick/Yardstick+Colorado+CI+Status Yardstick status (dfarrell07, 14:09:32)
    12. 4 successful builds in a row, need to track daily builds and fix bugs (acm, 14:10:30)
    13. Yardstick is working on moving to automatic reporting, vs updating this wiki, as functest demo'd recently (dfarrell07, 14:10:47)
    14. ACTION: morgan_orange to help yardstick with getting automatic test results (dfarrell07, 14:11:11)
    15. Maryam reports that VSPerf is quite green (dfarrell07, 14:11:59)
    16. yuyang reports bottlekneck green (dfarrell07, 14:12:25)
    17. mthahan some reds with intel pod going down over weekend (acm, 14:12:32)
    18. bottlenecks green (acm, 14:12:46)
    19. OAI had a demo at Summit (dfarrell07, 14:14:26)
    20. They have an open source VNF we are interested in integrating into functional testing (dfarrell07, 14:14:46)
    21. presented in berlin summit - not clear if integration possible, some licenses proprietary in open source VNF (acm, 14:15:49)
    22. (the usuall sharing issues) (acm, 14:16:16)
    23. There's a set of slides that we'll try to get a link to later (dfarrell07, 14:16:40)
    24. many use cases, including MEC (acm, 14:17:42)
    25. will send slides... (acm, 14:17:53)
    26. Already have various PoCs going on (dfarrell07, 14:18:43)
    27. Working with many partners (dfarrell07, 14:18:47)
    28. me too (acm, 14:19:16)
    29. eurocom is a major group, many members, Orange included (which is how we got linked I guess) (dfarrell07, 14:20:37)
    30. EURECOM - research and teaching, many affiliations. (acm, 14:20:47)
    31. yes (acm, 14:21:13)
    32. soft impl of eNodeB , other elements (acm, 14:22:04)
    33. Looking to work with more open source communities, like OPNFV (dfarrell07, 14:22:52)
    34. running on intel (acm, 14:22:53)
    35. IPR rich area - free for non-comercial use (acm, 14:23:24)
    36. RAN has license complexities, but is under Apache license with non-commercial clause (dfarrell07, 14:23:52)
    37. this should be disruptive for 5G (acm, 14:24:44)
    38. RF part - they have their own board, others are available. (acm, 14:25:31)
    39. Working with many different hardware projects/vendors (dfarrell07, 14:25:34)
    40. I have to go... (acm, 14:26:13)
    41. We're not lawyers, but there are efforts to make working with open source possible (dfarrell07, 14:27:12)
    42. OSA has big scope, but also quite focused on SDN 5G (dfarrell07, 14:28:21)
    43. Working with OpenStack (dfarrell07, 14:28:48)
    44. Interactions with ETSI, PoCs (dfarrell07, 14:29:44)
    45. Looking for thoughts about best way to use OAI as an upstream to test/integrate in OPNFV (dfarrell07, 14:31:52)
    46. Seem to be mostly looking at Fuel/JOID installers at the moment (dfarrell07, 14:33:02)
    47. morgan_orange +1s targeting D release to integrate a VNF into OPNFV (dfarrell07, 14:33:38)
    48. Basic steps 1) Deploy orchestrator 2) deploy VNF 3) develop tests to validate VNF 4) Get that running in CI so functest can validate (dfarrell07, 14:34:47)
    49. OPNFV is MANO, controller, installer agnostic, so should be able to get VNF deploying in various scenarios (dfarrell07, 14:35:41)
    50. Hardware and license are kinda hard with 5G space in general, morgan_orange says, but pushing to make it happen (dfarrell07, 14:36:31)
    51. First goal should be getting simple VNF like for vEPC (DB with algos to manage connectivity) running in community labs, which we can give you access to (dfarrell07, 14:37:46)
    52. Juju people have prio to show as VNFM for D (dfarrell07, 14:38:19)
    53. morgan_orange scopes this as possible for D (dfarrell07, 14:38:49)
    54. Discussion about how hardware isn't too bad to get, can be simulated (dfarrell07, 14:40:24)
    55. OSA folks need feedback about functest architecture, they will join weekly functest meeting for info, morgan_orange will also send pointers (dfarrell07, 14:41:10)
    56. http://artifacts.opnfv.org/functest/docs/configguide/ (jose_lausuch, 14:41:21)
    57. http://artifacts.opnfv.org/functest/docs/userguide/ (jose_lausuch, 14:41:32)
    58. OSA people will try to continue joining calls (dfarrell07, 14:42:02)
    59. http://artifacts.opnfv.org/functest/docs/userguide/ User guide for Functest (dfarrell07, 14:42:27)
    60. http://artifacts.opnfv.org/functest/docs/configguide/ config guide for functest (dfarrell07, 14:42:39)
    61. Discussion will continue in IRC-only functest meeting (dfarrell07, 14:43:18)
    62. Question about how to take part in community labs (dfarrell07, 14:43:59)
    63. "can we still work with you all" YES! :) (dfarrell07, 14:44:11)
    64. Al Morton back (acm, 14:44:19)
    65. OPNFV is an open source project, but more importantly has an open development model, you don't need to be a member to contribute (dfarrell07, 14:44:49)
    66. OPNFV membership not needed to work with us, have access to labs, with some variation across labs (acm, 14:45:09)
    67. we can give you pubkey ssh access to get into labs (dfarrell07, 14:45:13)
    68. create LF account for acces to other project aspects - Wikui, JIRA, etc. (acm, 14:46:24)
    69. ACTION: morgan to send link to docs about getting started as dev, IRC access, similar (dfarrell07, 14:47:05)
    70. https://wiki.opnfv.org/display/DEV/Developer+Getting+Started Overall OPNFV dev guide (dfarrell07, 14:47:58)
    71. http://webchat.freenode.net/?channels=opnfv-testperf (acm, 14:48:27)
    72. bryan_att asks if there's going to be hardware/software brought in that allows us to interact with OAI, to do some of this specialized stuff (dfarrell07, 14:50:11)
    73. OAI stuff can run in software (but with a specialized kernel), will be slow-ish but testing should be okay (dfarrell07, 14:51:41)
    74. Sounds like there will be changes to installers required, kernel mods/versions (dfarrell07, 14:52:55)
    75. there are still challenges to run OPEN Air on GP intel and ARM, (acm, 14:53:18)
    76. bryan_att asks if intent is to eventually add hardware to Pharos, Rohit says would rather do this in generic Intel/AMD (dfarrell07, 14:53:52)
    77. challenges to runt the code in KVM or Docker... (acm, 14:54:13)

  4. Discussions on Scenario generic/specific (dfarrell07, 14:55:03)
    1. morgan_orange talks about needed config per scenario (dfarrell07, 14:55:39)
    2. scenaio owners are the installers - (acm, 14:55:48)
    3. installers own laying down scenarios, but don't know this specific stuff, need to clarify how generic or specific logic is handled (dfarrell07, 14:56:33)
    4. morgan_orange believes this idea of specific scenarios makes sense. (acm, 14:57:45)
    5. (bryan_att) I meant the type of hardware that provides the access network interfaces for ON.Lab projects such as M-CORD (mobile) and R-CORD (optical residential). Augmenting OPNFV Pharos PODs with such hardware would enable OPNFV/CORD testing, e.g. such as would use the open air interface. (dfarrell07, 14:58:39)
    6. (bryan_att) But emulating those access network elements is also possible, though the performance will be impacted as noted. (dfarrell07, 14:58:51)
    7. Rohit: is it too soon to present this project to the TCS (heather and Ray suggested this as follow-up) (acm, 14:59:15)
    8. bryan: prefer the weekly tech discussion on thursday tech meeting, takes pressure off TSC (acm, 15:00:24)
    9. Discussion will continue on tech calls, mailing lists, clarifying and making progress, then after release may be better to bring to TSC (dfarrell07, 15:01:43)
    10. morgan_orange better to delay until late Sept October, after release. (acm, 15:01:49)


Meeting ended at 15:03:43 UTC (full logs).

Action items

  1. (carry-over, after release)kubi001 morgan_orange plan an update on grafana/ELK in September
  2. morgan_orange to help yardstick with getting automatic test results
  3. morgan to send link to docs about getting started as dev, IRC access, similar


Action items, by person

  1. kubi001
    1. (carry-over, after release)kubi001 morgan_orange plan an update on grafana/ELK in September
  2. morgan_orange
    1. (carry-over, after release)kubi001 morgan_orange plan an update on grafana/ELK in September
    2. morgan_orange to help yardstick with getting automatic test results


People present (lines said)

  1. dfarrell07 (75)
  2. acm (38)
  3. morgan_orange (4)
  4. collabot` (4)
  5. jose_lausuch (4)
  6. bryan_att (3)
  7. kubi001 (3)
  8. trevor_intel (1)
  9. yuyang (1)


Generated by MeetBot 0.1.4.