#fdio-csit: FD.io CSIT project meeting

Meeting started by mackonstan at 14:00:02 UTC (full logs).

Meeting summary

  1. Agenda bashing (mackonstan, 14:01:08)
    1. Tibor Frank (tifrank, 14:01:17)
    2. Dave Wallace (dwallacelf, 14:01:35)
    3. Vratko Polak (vrpolak, 14:02:29)
    4. Jan Gelety (jgelety, 14:02:45)

  2. Inputs from LFN and FD.io projects (mackonstan, 14:03:18)
    1. VPP - Dave updating (mackonstan, 14:03:30)
    2. v19.04.1 artifacts published last Monday (mackonstan, 14:03:44)
    3. can proceed with CSIT-1904_1 spotcheck testing (mackonstan, 14:04:17)
    4. more patches went into vpp stable/19.01 after v19.01.2 tag (mackonstan, 14:04:51)
    5. 1 patch in vpp stable/19.04 after v19.04.1 tag (mackonstan, 14:05:21)
    6. HC - Michal updating (mackonstan, 14:08:08)
    7. Not planning to migrate to vpp_device (mackonstan, 14:08:38)
    8. Currently no new features planned for HC that need to be tested (mackonstan, 14:09:11)
    9. Fine with VIRL shutdown - will manage testing within HC project (mackonstan, 14:10:10)

  3. Operational status (mackonstan, 14:10:39)
    1. Tom Herbert (tfherbert, 14:11:00)
    2. Ed updating - LF infra working (mackonstan, 14:11:07)
    3. Maciek updating - LF IT/Vexxhost confirmed s2-t12-sut1 (from 1n-skx testbed) is being replaced with spare, and faulty server RMAed (mackonstan, 14:12:35)
    4. Tom updating - Centos CI in vpp_device - patch submitted and rebased (mackonstan, 14:15:03)
    5. Peter comments - Tom's patch merged, some additional fixes applied (mackonstan, 14:15:37)
    6. Centos vpp_device tests working now (mackonstan, 14:15:52)
    7. next steps: test triggers, fix any issue (mackonstan, 14:17:16)
    8. once working then enable for semi-weekly and weekly execution (mackonstan, 14:17:40)
    9. then consider per patch (mackonstan, 14:17:57)
    10. VPP code performance (mackonstan, 14:18:22)

  4. VPP code performance (mackonstan, 14:18:38)
    1. Jan updating - currently perf mrr daily jobs taking >24hrs and queue is building up (mackonstan, 14:20:35)
    2. side effect of two things: suite generator and PAPI (mackonstan, 14:21:42)
    3. on average duration of MRR test increased from ~50sec to ~5min (mackonstan, 14:23:01)
    4. Jan update - it's 4-5 min per test case (mackonstan, 14:23:45)
    5. Jan update - the root cause for PAPI slowness is known - the issue is vpp.disconnect() where it is waiting util timout is finishs - it is 5s (mackonstan, 14:29:18)
    6. Jan update - and it is done per PapiExecutor session, it should be fixed with vpp socket connection (mackonstan, 14:30:08)
    7. Vratko - will test to see how much faster papi over unix remote socket is vs. current PAPI scheme (mackonstan, 14:30:53)
    8. Maciek - propose to cut the number of tests, by restricting them to specific NIC types: x520 and x710 to start with. (mackonstan, 14:35:58)
    9. Vratko - replace existing complex expression file, with a new one simple coarse filter matching on NIC types: x520 and x710 to start with. (mackonstan, 14:37:35)
    10. Jan - note that ipsec QAT HW will require xl710 - can be re-added later (mackonstan, 14:38:01)

  5. Releases (mackonstan, 14:38:28)
    1. CSIT-1901.2 report - published (mackonstan, 14:39:09)
    2. https://docs.fd.io/csit/rls1901_2/report/ (mackonstan, 14:39:20)
    3. CSIT-1904 report - Tibor update - more data added to the report, only 3n-hsw detailed tests (mackonstan, 14:40:28)
    4. https://docs.fd.io/csit/rls1904/report/introduction/report_history.html (mackonstan, 14:40:39)
    5. CSIT-1908 plan (mackonstan, 14:41:17)
    6. Rasti working on jira Epics for CSIT-1908 (mackonstan, 14:41:34)
    7. https://wiki.fd.io/view/CSIT/csit1908_plan (mackonstan, 14:42:45)

  6. Developments (mackonstan, 14:43:02)
  7. Installer (mackonstan, 14:43:45)
    1. Dnv onboarding - Peter helping Luyong to complete Dnv onboarding into jenkins (mackonstan, 14:44:11)

  8. Framework (mackonstan, 14:44:30)
    1. Tibor updating - VAT to PAPI - work progressing quite well, some issues with counters using VPPstats (mackonstan, 14:45:15)
    2. using cli and VPPstats produces different results - being investigated (mackonstan, 14:46:39)
    3. Jan working on IP libraries - some refactoring required (mackonstan, 14:47:03)

  9. VPP-API "Flag Day" (mackonstan, 14:48:02)
    1. Dave - after review with Damjan had to abandon the vpp patch tracking api changes. Updating the document to reflect that. (mackonstan, 14:50:03)
    2. Dave - looking to switch from testing and storing api files, and tracking CRCs per message. But these in turn are not capturing complete api msgs including the metadata. (mackonstan, 14:50:50)
    3. Dave - expecting the next rev of the draft today/tomorrow (mackonstan, 14:53:20)
    4. Vratko - will adjust csit side implementation accordingly. (mackonstan, 14:54:21)

  10. Test environments (mackonstan, 14:54:30)
    1. Sirshak - confirmed Juraj should be joining next week, so work on Arm testbeds should restart (mackonstan, 14:55:32)

  11. New tests (mackonstan, 14:55:56)
    1. IPsec - Peter - redone IPsec tests, now properly scaling up with multi-core (mackonstan, 14:56:31)

  12. TRex (mackonstan, 14:56:46)
    1. Peter - v2.54 is doing very well in production, stable (mackonstan, 14:57:05)

  13. FD.io website (mackonstan, 14:59:31)
    1. moving to github driven web page (mackonstan, 14:59:43)
    2. TRex (mackonstan, 14:59:46)
    3. https://myfdio.netlify.com/projects/ (mackonstan, 14:59:56)
    4. https://github.com/jadenisco/fdiomain (mackonstan, 15:00:21)
    5. above links are for sandbox (mackonstan, 15:00:27)


Meeting ended at 15:02:32 UTC (full logs).

Action items

  1. (none)


People present (lines said)

  1. mackonstan (67)
  2. collabot` (4)
  3. vrpolak (1)
  4. dwallacelf (1)
  5. jgelety (1)
  6. tfherbert (1)
  7. tifrank (1)


Generated by MeetBot 0.1.4.