#fdio-meeting: FD.io CSIT Project Meeting

Meeting started by mackonstan at 15:01:54 UTC (full logs).

Meeting summary

    1. mackonstan (mackonstan, 15:02:22)
    2. Dave Wallace (dwallacelf, 15:02:51)

  2. Agenda bashing (mackonstan, 15:02:57)
    1. viliam luc (viliamluc, 15:03:52)
    2. Juraj Linkeš (jlinkes, 15:03:54)

  3. VPP updates (mackonstan, 15:04:39)
    1. Andrew - vpp community meeting yesterday, Dave showed auto-abandon stuff clearing the untouched backlog, discussed defering dpdk bumping (mackonstan, 15:06:02)
    2. Andrew - apart from Andrew and Dave, Fan was taking part in the discussion (Kai couldn't join the vpp call yesterday), but Kai agrees with the decision to defer dpdk bumping (mackonstan, 15:09:51)
    3. mackonstan - going forward we are adding more drivers (dpdk mlx) and more brand new cutting edge hw (spr, pcie gen5, mlx cx6, mlx cx7 2p200ge) (mackonstan, 15:18:41)

  4. TSC updates (mackonstan, 15:19:02)
    1. Dave - creating FD.io webinar schedule - updated fd.io wiki with the new link pointing to webinar schedule (mackonstan, 15:19:33)

  5. News (mackonstan, 15:21:16)
    1. Maciek - it is in public domain that VPP can do ~2tbps ipsec on 2 socket server - spr server (mackonstan, 15:21:49)
    2. https://networkbuilders.intel.com/solutionslibrary/intel-avx-512-high-performance-ipsec-with-4th-gen-intel-xeon-scalable-processor-technology-guide (mackonstan, 15:21:52)

  6. E810 flow "offload" tests (mackonstan, 15:25:23)
    1. Kai - discussing the root cause(s) behind different performance for gtpusw / gtpuhw - looks like it's down to HW differences between the two perf test environments (mackonstan, 15:28:24)

  7. Scale tests API optimization (mackonstan, 15:28:50)
    1. Vratko - relevant changes merged - so far rc1 results looking good, waiting for more data trending, looks like it is working as expected (mackonstan, 15:30:01)

  8. Next release CSIT-2302 (mackonstan, 15:31:29)
    1. Tibor (from offline report) - All RC1 runs done, All TRex runs done, All DPDK runs done (mackonstan, 15:32:12)
    2. RC1 report generated: https://s3-docs.fd.io/csit/master/report/index.html (mackonstan, 15:32:29)
    3. https://s3-docs.fd.io/csit/master/report/index.html (mackonstan, 15:32:41)
    4. docs.fd.io/csit/master/report/index.html (mackonstan, 15:33:28)
    5. updated data in the report: All iterative data is RC1; Ignore coverage data; Graphs, comparison tables - done, up-to-date; To be added: Comparison tables for TRex (mackonstan, 15:34:27)
    6. mackonstan - re hoststack tests - all are working, but quic; nginx ldpreload are using older version of nginx; (mackonstan, 15:36:05)
    7. dave - re quic - reproduced the crash issue, working on rca'ing it (mackonstan, 15:37:16)
    8. dave - re quick - wip, no line of sight yet, labour intensive (mackonstan, 15:37:41)
    9. mackonstan - my understanding is that mlx dpdk tests are coded but still need to run (mackonstan, 15:40:26)
    10. kai - re wireguard 10k tunnels - tests are working but code not merged as still working thru comments to vpp patch from wireguard maintainer (mackonstan, 15:41:58)
    11. mackonstan - note on rc2 date - 8-feb (mackonstan, 15:42:31)
    12. https://wiki.fd.io/view/CSIT/csit2302_plan (mackonstan, 15:43:27)

  9. Physical Infrastructure (mackonstan, 15:47:25)
    1. mackonstan - looks like there is a customs bill for mlx nics that arrived from israel (mackonstan, 15:47:47)
    2. dave - will discuss it at tomorrow's FD.io TSC meeting (mackonstan, 15:48:05)

  10. Virtual Infrastructure (mackonstan, 15:48:25)
    1. Dave - ci-mgmt branch was broken last week, following updates / changes to LFN global jjb version (mackonstan, 15:49:25)
    2. Dave - fixed now (mackonstan, 15:49:34)

  11. Trending and Test Failures (mackonstan, 15:49:52)
    1. Viliam - 3n-alt interfaces sometimes are not coming up see item 3.1.1 in tracking wiki, link below (mackonstan, 15:50:55)
    2. https://wiki.fd.io/view/CSIT/TestFailuresTracking (mackonstan, 15:51:09)
    3. testpmd/l3fwd issues: https://jira.fd.io/browse/CSIT-1848 (jlinkes, 15:54:26)

  12. mlx nics for altra server (mackonstan, 15:58:34)
    1. see juraj's email about what mlx nics to put into altra, cx6 and cx7 (mackonstan, 15:58:57)
    2. juraj - need to also account for nics going into trex (mackonstan, 15:59:11)


Meeting ended at 15:59:17 UTC (full logs).

Action items

  1. (none)


People present (lines said)

  1. mackonstan (43)
  2. collab-meetbot` (10)
  3. jlinkes (2)
  4. dwallacelf (1)
  5. viliamluc (1)
  6. vrpolak (0)
  7. tifrank (0)


Generated by MeetBot 0.1.4.