#fdio-csit: FD.io CSIT project meeting

Meeting started by mackonstan at 13:59:03 UTC (full logs).

Meeting summary

  1. Agenda bashing (mackonstan, 14:00:57)
    1. Juraj Linkeš (jlinkes, 14:02:27)

  2. FD.io CSIT physical labs (mackonstan, 14:02:43)
    1. Dave Wallace (dwallacelf, 14:03:25)
    2. FD.io lab extension - agree over email re setup per meeting agenda (mackonstan, 14:04:04)

  3. Inputs from LFN and FD.io projects (mackonstan, 14:04:22)
    1. VPP v19.08 content updated on the wiki (mackonstan, 14:04:40)
    2. https://wiki.fd.io/view/Projects/vpp/Release_Plans/Release_Plan_19.08 (mackonstan, 14:05:15)
    3. Dave Wallace: vpp maintenance releases schedule 17-Jul 19.01.3, 24-Jul 19.04.2 (mackonstan, 14:06:07)
    4. CSIT should plan for executing set of tests per usual maintenance practice (mackonstan, 14:06:47)
    5. Jan Gelety (jgelety, 14:07:23)
    6. last VPP call: vpp-dev folks confirmed they will be notifying csit-dev list if/when patches with changed api's are expected to show up (mackonstan, 14:10:30)

  4. HoneyComb - no updates (mackonstan, 14:10:51)
  5. TSC (mackonstan, 14:11:06)
    1. no meeting last week (mackonstan, 14:11:14)

  6. Operational status (mackonstan, 14:11:24)
    1. few tests failing in perf-mrr-daily: ipsec and x520 due to papi failures, memif on hsw unclear, vic due to 9200 mtu set failing (mackonstan, 14:14:46)
    2. Vratko and Peter on top of those failures (mackonstan, 14:15:04)

  7. Centos CI (mackonstan, 14:15:22)
    1. Peter: Centos job prepared for vpp_Device, need Ed to have it fully onboarded (mackonstan, 14:15:42)
    2. related to vpp_device Juraj : arm onboarding into vpp_Device - progressing well, most patches already in review (mackonstan, 14:18:55)
    3. Juraj: major open point - automated pipeline for posting images to dockerhub for both x86 and arm (mackonstan, 14:21:43)
    4. Juraj: only one Arm server is available for vpp_device today. Not suitable for production with voting, but would still like to have a Jenkins job to run tests. Steps being taken by Arm team to get another server. (mackonstan, 14:23:35)

  8. VPP code performance (mackonstan, 14:23:59)
    1. ip4 and ip6 tests back to previous performance - past regression caused by programming recursive routes (mackonstan, 14:27:11)

  9. Developments Framework (mackonstan, 14:27:47)
    1. Jan: VAT to PAPI - ipsec util wip, lisp util wip (mackonstan, 14:28:57)
    2. PAPI not good for scale test, sticking to VAT (mackonstan, 14:30:05)
    3. PAPI over remote socket - Vratko: not fast enough for scale tests, hoping to finish the main patch set by tomorrow, merge by eow, looking for more reviews (mackonstan, 14:32:48)
    4. (mackonstan, 14:34:35)
    5. Vratko: VPP-API "Flag Day" gate-keeping - once done with remote socket, will create a new job for vpp-csit-api check (mackonstan, 14:37:08)
    6. https://gerrit.fd.io/r/19272 <-- main socket papi change (part 2) (vrpolak, 14:38:55)
    7. https://gerrit.fd.io/r/20170 <-- csit preparation before main socket papi (part 1) (vrpolak, 14:39:47)
    8. Peter: testing_path and performance_setup L1 KWs alignment between -func/-dev and -perf (mackonstan, 14:41:16)
    9. Peter: aligned majority of the KWs, -dev and -perf tests aligned (mackonstan, 14:41:41)
    10. Peter: now picking up any remaining virl -func tests for running them in -dev (mackonstan, 14:42:53)
    11. Peter: will upload the xls to google docs for csit-dev review (mackonstan, 14:43:27)

  10. Test environments (mackonstan, 14:43:45)
    1. Peter: pile of changes from Juraj to onboard arm to vpp_device (mackonstan, 14:44:08)
    2. New tests (mackonstan, 14:45:01)

  11. New tests (mackonstan, 14:45:04)
    1. Peter: IPsec scale for 60k tunnel is extremely slow, config speed issue identified with vpp-dev/ Neale, fix in progress (mackonstan, 14:45:41)
    2. Peter: 2-node vxlan vhost-VM tests progressing (mackonstan, 14:46:01)
    3. Peter: adding AVF driver tests to vpp_device, some issue with dot1q processing (mackonstan, 14:46:23)

  12. TRex (mackonstan, 14:46:55)
    1. Latency HdrHistogram - added to latest TRex version (mackonstan, 14:47:21)
    2. Vratko: looking at it and see how hard/easy to use (mackonstan, 14:49:30)


Meeting ended at 14:52:36 UTC (full logs).

Action items

  1. (none)


People present (lines said)

  1. mackonstan (45)
  2. collabot` (4)
  3. vrpolak (2)
  4. jgelety (1)
  5. jlinkes (1)
  6. dwallacelf (1)


Generated by MeetBot 0.1.4.