#fdio-csit: FD.io CSIT Project Meeting

Meeting started by mackonstan at 14:00:42 UTC (full logs).

Meeting summary

    1. Jan Gelety (jgelety, 14:00:46)
    2. Tibor Frank (tifrank, 14:00:47)

  2. Agenda bashing (mackonstan, 14:01:45)
  3. Inputs from projects - VPP (mackonstan, 14:03:15)
    1. Dave, jenkins ui slow/unavailable, Vanessa confirmed likely hitting gerrit-plugin bug (mackonstan, 14:04:29)
    2. Tibor, current ~10 performance jobs using testbeds, another 40 in-waiting. (mackonstan, 14:06:05)
    3. Dave, the only known solution is to apply the fix and reload Jenkins. (mackonstan, 14:08:17)
    4. Tibor, confirmed now is as good time as any, so green light to proceed with Jenkins emergency maintenance. (mackonstan, 14:08:51)
    5. Dave, will work with Vanessa (mackonstan, 14:08:59)

  4. Inputs from projects - VPP (mackonstan, 14:09:33)
    1. Inputs from projects - TSC (mackonstan, 14:10:12)

  5. Releases - CSIT-2009 (mackonstan, 14:10:39)
    1. Tibor, performance jobs executing successfully so far, graphs already generated from iterative tests (now just adding runs) (mackonstan, 14:11:52)
    2. Tibor, started coverage tests too, nfv_density, reconfig, soak progressing too (mackonstan, 14:12:15)
    3. Tibor, comparison tables are also generated, RCA for progressions/regressions to be still processed (mackonstan, 14:13:21)
    4. Vratko, Maciek, look at RCAing the comparisons, once majority of runs are completed (skx already 10runs, clx completed 6 runs) (mackonstan, 14:15:04)
    5. Maciek, report, working on methodology sections, adding nat44, editing ip4. (mackonstan, 14:17:58)
    6. Jan, added rls notes for vpp_Device. (mackonstan, 14:18:25)
    7. Maciek, Jan, will update rls notes for other sections. (mackonstan, 14:18:36)
    8. Vratko, new stateful nat44ed cps and pps tests, NDR PDR tests are (almost!) working, but issues with TRex MRR tests due to inconsistent TRex behaviour. (mackonstan, 14:23:50)
    9. Jan, Peter, nat44det and nat44ed unidir using TRex STL tests are running, the only anomaly is MRR value is quite higher than PDR and NDR. (mackonstan, 14:28:08)

  6. Physical Infrastructure (mackonstan, 14:28:23)
    1. Two skx testbeds and one tishan were down last Friday, now back, but no note in the ticket (mackonstan, 14:30:42)
    2. Juraj, confirmed there was some communication from Vexxhost about uplink cable problem (mackonstan, 14:31:07)
    3. https://secure.vexxhost.com/billing/viewticket.php?tid=WEX-010067&c=c4eZTlM8 (mackonstan, 14:32:19)
    4. Juraj, new ThunderX2s coming online for vpp_Device (mackonstan, 14:32:50)
    5. Juraj, looking to use the other ThunderX2 for perf testing (mackonstan, 14:33:20)
    6. Juraj, looking to share the Xeon-Skx/TG server between existing Taishan and new ThunderX2. (mackonstan, 14:34:52)

  7. Keesang, AMD, sent lstopo info with numa node0,1 mappings to PCIe devices. (mackonstan, 14:36:23)
    1. Keesang, explained the AMD zen2 cpu numa configuration. (mackonstan, 14:39:02)
    2. Keesang, both servers should be identical, will reconfirm. (mackonstan, 14:39:31)
    3. Peter, few more things to do for AMDs: Linux params tuning, dry runs. (mackonstan, 14:40:45)
    4. Keesang, Peter, Maciek to coordinate onboarding. (mackonstan, 14:41:04)
    5. Maciek, discussion ongoing about 2021 lab extension (mackonstan, 14:43:14)

  8. Virtual Infrastructure, CI/CD (mackonstan, 14:45:22)
    1. Dave, Vanessa proceeding with the emergency maintenance right now (bringing forward complete maintenance originally scheduled for 19-Oct) (mackonstan, 14:47:45)
    2. Jan, all trending jobs were affected by VPP getstats bug, now fixed by Ole (mackonstan, 14:48:32)
    3. Dave, not planning any new TWS call for time being (mackonstan, 14:49:02)
    4. Dave, finishing docker images for build jobs for Centos8, Debian10 for v20.09 (mackonstan, 14:50:21)
    5. Dave, not touching images for CSIT jobs (mackonstan, 14:50:35)

  9. Environment, Framework (mackonstan, 14:50:52)
  10. Performance Tests (mackonstan, 14:51:02)
    1. Jan, all trending tests were affected by VPP getstats bug, now fixed by Ole. (mackonstan, 14:56:03)
    2. https://gerrit.fd.io/r/c/vpp/+/29114 (mackonstan, 14:56:16)
    3. Andrew, this patch should not have passed make test, need to plug this test escape (mackonstan, 14:56:38)
    4. Jan, propose to add it also to vpp_device (mackonstan, 14:57:05)
    5. Andrew, we now have daily rls notes, automatically generated, based on parsing commits type: feature. (mackonstan, 15:00:08)
    6. https://github.com/vpp-dev/vpp-release-notes/blob/master/RELEASE.md (mackonstan, 15:00:13)


Meeting ended at 15:01:01 UTC (full logs).

Action items

  1. (none)


People present (lines said)

  1. mackonstan (50)
  2. collab-meetbot (4)
  3. jgelety (1)
  4. tifrank (1)


Generated by MeetBot 0.1.4.