====================================== #fdio-csit: FD.io CSIT project meeting ====================================== Meeting started by mackonstan at 15:05:35 UTC. The full logs are available at http://ircbot.wl.linuxfoundation.org/meetings/fdio-csit/2020/fdio-csit.2020-03-04-15.05.log.html . Meeting summary --------------- * Jan Gelety (jgelety, 15:05:40) * Tibor Frank (tifrank, 15:05:41) * Dave Wallace (dwallacelf, 15:05:43) * Vratko Polak. (vrpolak, 15:05:44) * Juraj Linkeš (jlinkes, 15:05:54) * Agenda bashing (mackonstan, 15:06:04) * DST starts (mackonstan, 15:08:32) * keeping meeting time same for Europe/UK, PDT/EDT guys have it one hour later (local time). (mackonstan, 15:09:16) * Physical labs, testbeds (mackonstan, 15:09:28) * Peter update - all Skx machines upgraded to BIOS v3.2 per Intel recommendations to address ucode issues; SSD firmware to avoid RMAs leading to some SSDs dying and needed to be replaced and OS reinstalled. (mackonstan, 15:12:21) * All skx testbeds are now in operation. (mackonstan, 15:12:34) * Govind from Arm is still waiting for VPN access, ticket open. (mackonstan, 15:14:58) * Inputs from projects (mackonstan, 15:15:27) * Dave update: VPP - v20.05 schedule F0 06-May (mackonstan, 15:17:49) * Maciek update: TSC - Ray presented Scapy/GPLv2 points, TSC voting expected this or next week. (mackonstan, 15:19:38) * Maciek update: TSC - vpp dev event collocated with dpdk dev event in France in october (mackonstan, 15:20:33) * Releases CSIT-2001 (mackonstan, 15:21:05) * Testing - Jan update - last ndrpdr select test running for Tsh, all other performance tests are done. (mackonstan, 15:22:05) * Testing - Jan - soak 2n-clx running (some vhost test failures), reconfig 2n-clx done, 2n-clx and 3n-hsw hoststack done (mackonstan, 15:24:13) * Testing - Jan - only Skx tests left to close CSIT-2001. Trending cadence can be returned to normal for all but skx testbeds. (mackonstan, 15:25:08) * Report - Tibor - latest maintenance version is out, all produced test data included: (mackonstan, 15:26:51) * LINK: https://docs.fd.io/csit/rls2001/report/ (mackonstan, 15:27:00) * additions are recorded here: (mackonstan, 15:27:32) * LINK: https://docs.fd.io/csit/rls2001/report/introduction/report_history.html (mackonstan, 15:27:34) * ACTION: Dave, Florin, Maciek - would be good to review the hoststack testing sections (mackonstan, 15:31:23) * LINK: https://docs.fd.io/csit/rls2001/report/vpp_performance_tests/hoststack_testing/index.html (mackonstan, 15:31:36) * Skx update - Maciek, Peter - BIOS upgraded to v3.2, very low NDR issue still open on all but one 2n-skx testbed. increasing NIC driver ring size elevates the issue. Under investigation with Intel, another sync tomorrow. If not line of sight to resolution, looking to go ahead with CSIT-2001 testing with increased ring sizes and compare to default setting. (mackonstan, 15:43:43) * Releases CSIT-1908 (mackonstan, 15:45:54) * Peter - some fixes applied, vpp-device tests are now running. (mackonstan, 15:46:16) * Vratko - like more changes required for performance tests. (mackonstan, 15:46:38) * Dave - will ask on the next VPP call, if any calls for v19.08.2 rls (mackonstan, 15:49:36) * Operations (mackonstan, 15:49:49) * Ed Kern - all quiet, all working (mackonstan, 15:50:08) * Ed Kern - disabling retries - patch open in ci-mgmt (mackonstan, 15:50:43) * vpp-csit-device tests - re-enable voting - Vratko - infrequent failures still happening, needs to be investigate (mackonstan, 15:51:47) * Ed - while Vanessa away, stand in is eball@linuxfoundation.org (mackonstan, 15:52:20) * Centos CI - no update (mackonstan, 15:52:33) * VPP code performance (mackonstan, 15:55:47) * number of issues in trending in master - see email summaries in csit-report mailer (mackonstan, 15:56:28) * Developments (mackonstan, 15:56:44) * API functional test coverage - Vijay - IPsec tests, encryption works decryption doesn't (mackonstan, 15:57:31) * Performance: No new regressions. I tried to bisect for two new failures on 3n-hsw, but results were not stable, the failure is not deterministic. Finally, there is one apparent progression on 3n-hsw I tracked to one testbed having better performance. Examining it now. (vrpolak, 15:59:26) Meeting ended at 16:01:30 UTC. People present (lines said) --------------------------- * mackonstan (40) * collabot` (4) * vrpolak (2) * jgelety (1) * dwallacelf (1) * jlinkes (1) * tifrank (1) Generated by `MeetBot`_ 0.1.4