====================================== #fdio-csit: FD.io CSIT project meeting ====================================== Meeting started by mackonstan at 13:59:37 UTC. The full logs are available at http://ircbot.wl.linuxfoundation.org/meetings/fdio-csit/2019/fdio-csit.2019-06-12-13.59.log.html . Meeting summary --------------- * Agenda bashing (mackonstan, 14:01:43) * Vratko Polak (vrpolak, 14:02:16) * Jan Gelety (jgelety, 14:02:29) * Juraj Linkeš (jlinkes, 14:02:32) * Dave Wallace (dwallacelf, 14:02:36) * FD.io CSIT physical labs - Xeon testbeds (mackonstan, 14:05:06) * faulty server for vpp_device replaced and onboarded - thx to vexxhost, Peter, Ed! (mackonstan, 14:06:16) * Ed confirmed it's in production for vpp-device jobs (mackonstan, 14:07:59) * Dave: number of failures observed in vpp-csit-verify-device (mackonstan, 14:09:15) * Peter: newly added ipsec tests failing, due to vpp api/code change (mackonstan, 14:10:17) * Need to enable voting for vpp-csit-verify-device, so let's disable failing ipsec tests, get them fixed, and then re-add to the job once working (mackonstan, 14:21:55) * Arm testbeds (mackonstan, 14:23:27) * Atom testbeds (mackonstan, 14:23:33) * Peter: helping Luyong to onboard Atom perf tests onto FD.io jenkins (mackonstan, 14:24:11) * Inputs from LFN and FD.io projects (mackonstan, 14:24:23) * VPP (mackonstan, 14:24:36) * Dave: vpp maintenance releases out, v19.01.2 and v19.04.1, few patches added to respective stable branches, no calls for new maintenance releases (mackonstan, 14:25:32) * Dave: 1908 plan is published (mackonstan, 14:25:40) * HoneyComb (mackonstan, 14:26:00) * no updates (mackonstan, 14:26:36) * TSC (mackonstan, 14:27:00) * Vratko: discussion about updating the new FD.io webpage, responsibility lines between mktg committee and projects, there will be a presentation coving that (mackonstan, 14:27:59) * Operational status (mackonstan, 14:28:28) * no issues (mackonstan, 14:28:38) * Centos CI (mackonstan, 14:29:00) * no update (mackonstan, 14:29:03) * VPP code performance (mackonstan, 14:29:16) * Performance trending - Vratko: reduced daily trending to x710 NICs only, now daily jobs take <6hrs each (mackonstan, 14:30:14) * Vratko: next step is to add remaining NICs for base tests only, no features, see how much time (mackonstan, 14:31:25) * Maciek: proposal - let's review what is actually test, then add scale for x710, followed by base (no features) on other NICs (mackonstan, 14:33:19) * Releases (mackonstan, 14:35:37) * CSIT-1904 report, with more detailed results from 3n-hsw out today, and the final one next week. (mackonstan, 14:37:49) * CSIT-1904_1 report out later today (mackonstan, 14:38:03) * CSIT-1908 plan (mackonstan, 14:41:15) * LINK: https://wiki.fd.io/view/CSIT/csit1908_plan (mackonstan, 14:41:23) * Developments - Installer (mackonstan, 14:41:33) * readme file is uptodate covering: 3n, 2n, 1n (part) (mackonstan, 14:42:39) * Yulong used it for denverton and it worked (mackonstan, 14:42:50) * Dave: will use it for testing own dev and perf testbed (mackonstan, 14:43:22) * Dave: added a patch for interface discovery: Add interface discovery to update_topology script. (mackonstan, 14:44:20) * Dave: added a patch for interface discovery: https://gerrit.fd.io/r/c/20077/ (mackonstan, 14:44:50) * LINK: https://gerrit.fd.io/r/c/20077/ (mackonstan, 14:46:06) * Dave: wiki content outdated, we should refresh, especially a quick start guide (mackonstan, 14:47:05) * Peter: suggest we stick to keeping uptodate content in repo, and point from wiki (mackonstan, 14:47:24) * Peter: place it in readme, e.g. the one in the main csit directory (mackonstan, 14:48:00) * LINK: https://github.com/FDio/csit (mackonstan, 14:48:08) * Framework - VAT to PAPI migration (mackonstan, 14:49:26) * Tibor: things are going okay so far, lots of patches getting merged (mackonstan, 14:49:42) * Tibor: one issue - "show run" PAPI reporting different data compared to VAT/CLI (mackonstan, 14:50:25) * Vratko: two issues affecting performance: i) delay between disconnect and connect - fixed with remote socket, ii) need to support async calls so that multiple api calls can proceed in parallel (mackonstan, 14:52:34) * Vratko: PAPI over remote socket - looking good, one issue blocking merge - when vpp restarts new socket init works, but it doesn't work between suites (mackonstan, 14:54:10) * Framework - L1KWs alignment -func/-dev and -perf (mackonstan, 14:55:04) * Peter: need to align two files so that going forward any -perf test suite can be easily ported to -dev (mackonstan, 14:55:54) * LINK: https://git.fd.io/csit/tree/resources/libraries/robot/shared/testing_path.robot (mackonstan, 14:57:20) * LINK: https://git.fd.io/csit/tree/resources/libraries/robot/performance/performance_setup.robot (mackonstan, 14:57:33) * Framework - VPP-API "Flag Day" gate-keeping (mackonstan, 14:57:56) * Dave: adjusting to per api msg checks (mackonstan, 14:58:19) * Vratko: working on updated code, should know in next few days if the approach works :) (mackonstan, 15:00:41) * New tests (mackonstan, 15:02:24) * AVF driver test coverage in vpp_device - Vratko: discussing with Peter what the best approach (mackonstan, 15:03:02) Meeting ended at 15:03:05 UTC. People present (lines said) --------------------------- * mackonstan (58) * collabot` (3) * vrpolak (1) * jgelety (1) * dwallacelf (1) * jlinkes (1) Generated by `MeetBot`_ 0.1.4