====================================== #fdio-csit: FD.io CSIT project meeting ====================================== Meeting started by mackonstan at 14:00:39 UTC. The full logs are available at http://ircbot.wl.linuxfoundation.org/meetings/fdio-csit/2019/fdio-csit.2019-04-24-14.00.log.html . Meeting summary --------------- * Agenda bashing (mackonstan, 14:02:17) * Jan Gelety (jgelety, 14:02:46) * Tom Herbert (tfherbert, 14:03:46) * Inputs from LFN and FD.io projects (mackonstan, 14:04:06) * VPP (mackonstan, 14:04:11) * Peter Mikus (pmikus, 14:04:41) * Vratko Polak (vrpolak, 14:04:51) * DaveW - vpp-19.04 rls day today. All prepared to have the rls artefacts later today. (mackonstan, 14:05:00) * HoneyComb (mackonstan, 14:05:32) * Michal - RC1 last week, jvpp, hc, hc2vpp - all stable branches cut, but issues with Jenkins, some artefacts not uploaded (mackonstan, 14:06:52) * Dave Wallace (dwallacelf, 14:07:11) * Michal - 19.04 release data is 30-Apr. (mackonstan, 14:07:40) * TSC (mackonstan, 14:08:21) * no updates (mackonstan, 14:08:25) * Operational status (mackonstan, 14:09:40) * trending jobs failing - caused by API change, Vratko patched, should be good now. (mackonstan, 14:10:31) * Centos CI testing status - Tom Herbert - starting on it today, hoping to have it done today, requires fix in vpp spec files in vpp repo, followed by vpp_device updated for Centos (mackonstan, 14:12:20) * VPP code performance (mackonstan, 14:13:03) * buffer allocation issue with multi-core scale tests - root cause identified and configuration fix implemented. (mackonstan, 14:15:58) * LINK: https://gerrit.fd.io/r/#/c/18743/ (mackonstan, 14:16:10) * root cause is DPDK restriction to total of 256 memory mappings for buffer-to-hugepage (mackonstan, 14:20:26) * solution is to set buffers-per-numa to the maximum value supported by DPDK, now known. (mackonstan, 14:21:34) * IPsec "interface mode" tests are failing due to changes in code and moving logic into external plugin, this was fixed on CSIT side, but unfortunatelly AES-GBC is still not working (pmikus, 14:22:11) * another issue is with ipsec tunnel-interfaces (mackonstan, 14:22:46) * https://gerrit.fd.io/r/#/c/19131/ is fix from Neale to try to fix AES-GBC but still not working on 100% (pmikus, 14:23:00) * vpp code resolution is still in progress (mackonstan, 14:23:11) * LINK: https://gerrit.fd.io/r/#/c/19131/ (mackonstan, 14:23:21) * unlikely to target 19.04 rather 19.04.1 (pmikus, 14:23:22) * 3rd issue - is with link bonding tests: i) IMIX tests crashing vpp, ii) 2-link bond tests not working. (mackonstan, 14:26:22) * Release 19.04 (mackonstan, 14:26:50) * 19.04 rc2 dry runs - JanG - 4th issue: memif interfaces not up in subset of tests - need to investigate. (mackonstan, 14:28:35) * vpp-per-patch - Vratko - merged. Email to follow. (mackonstan, 14:30:01) * PLRsearch soak tests - Vratko - results stable, but setup sometimes fails, t/shooting in progress. 2hrs run set as default. Need to select few suites for inclusion into report, comparing results vs. NDRPDR. (mackonstan, 14:32:16) * NFV density - Peter - addressing comments, implementation completed, need to agree on naming for new tests. (mackonstan, 14:33:02) * 1904 report - Tibor - draft is out, includes test data from rc2 dry-runs (mackonstan, 14:34:55) * LINK: https://docs.fd.io/csit/master/report/vpp_performance_tests/packet_throughput_graphs/l2-3n-hsw-x520.html#b-1t1c-base-and-scale (mackonstan, 14:35:04) * see Tibor's email re report TODOs (mackonstan, 14:38:45) * Developments (mackonstan, 14:40:22) * automate VPP-API "Flag Day" (mackonstan, 14:41:28) * Vratko - implemented feature to check for API change - storing papi .json files and comparing them with the ones generated from vpp compilation (mackonstan, 14:43:01) * Vratko - also updated bootstrap script (mackonstan, 14:46:47) * LINK: https://gerrit.fd.io/r/#/c/19027/ (mackonstan, 14:46:50) * can not be merged until all things are in place for API flag day (mackonstan, 14:48:10) * Maintenance releases (mackonstan, 14:51:55) * DaveW - which should come first 19.01.2 or 19.04.1 ? (mackonstan, 14:52:22) * DaveW - discuss on VPP community call next week. (mackonstan, 14:56:34) Meeting ended at 14:56:44 UTC. People present (lines said) --------------------------- * mackonstan (41) * pmikus (5) * collabot (4) * vrpolak (1) * jgelety (1) * dwallacelf (1) * tfherbert (1) Generated by `MeetBot`_ 0.1.4