15:03:23 <mackonstan> #startmeeting FD.io CSIT project meeting 15:03:23 <collab-meetbot`> Meeting started Wed Feb 8 15:03:23 2023 UTC and is due to finish in 60 minutes. The chair is mackonstan. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:03:23 <collab-meetbot`> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:03:23 <collab-meetbot`> The meeting name has been set to 'fd_io_csit_project_meeting' 15:03:56 <tifrank> #info Tibor Frank 15:05:39 <mackonstan> #info mackonstan 15:05:50 <mackonstan> #chair dwallacelf jlinkes vrpolak mackonstan tifrank 15:05:50 <collab-meetbot`> Warning: Nick not in channel: dwallacelf 15:05:50 <collab-meetbot`> Warning: Nick not in channel: jlinkes 15:05:50 <collab-meetbot`> Warning: Nick not in channel: vrpolak 15:05:50 <collab-meetbot`> Warning: Nick not in channel: mackonstan 15:05:50 <collab-meetbot`> Warning: Nick not in channel: tifrank 15:05:50 <collab-meetbot`> Current chairs: dwallacelf jlinkes mackonstan tifrank vrpolak 15:06:06 <vrpolak> #info Vratko Polak. 15:06:20 <mackonstan> #topic Agenda bashing 15:08:32 <mackonstan> #info Peter - work in progress re Georgii's request for job_specs conversion into robot, but Georgii is away from office this week 15:08:59 <mackonstan> #topic Inputs from projects - VPP 15:09:14 <mackonstan> #info Tibor - didn't hear from Andrew re rc2 status 15:11:21 <mackonstan> #info Tibor - confirmed rc2 artefacts are in package cloud, pushed one hr ago 15:11:35 <mackonstan> #topic Inputs from projects - TSC 15:13:02 <mackonstan> #info mackonstan - re decommissioning of old equipment from lab, Casey confirmed that LFN team is okay with donating stuff that is not needed and re-confirmed that TSC voting is enough for approval. Casey emphasized that s/n records must be update for inventory tracking purposes. 15:14:24 <mackonstan> #info Dave - LFN internships are here - proposal window open till beginning of March 2023 15:15:33 <mackonstan> #topic E810 flow "offload" tests 15:16:20 <mackonstan> #info Kai - email conversation between Yulong and Vratko - see email thread "[csit-dev] CSIT E810 flow offload perf difference between Intel Lab" 15:16:21 <dwallacelf> #link https://wiki.lfnetworking.org/display/LN/LFN+Mentorship+Program 15:16:57 <jlinkes> #info Juraj Linkeš 15:17:00 <mackonstan> #info Kai - Yulong is looking to match csit hw setup and environment in his local labs 15:17:41 <mackonstan> #topic CSIT-2302 15:20:38 <mackonstan> #info mackonstan - added 100ge e810cq tests to job_specs for release as they were missing - 38165: job_specs: add e810cq 100ge tests to 2n-icx and 3n-icx | https://gerrit.fd.io/r/c/csit/+/38165 15:20:48 <mackonstan> #link https://gerrit.fd.io/r/c/csit/+/38165 15:21:54 <mackonstan> #info Tibor - merged it already, the only concern is the job duration for rls testing and for trending 15:23:09 <mackonstan> #info Tibor - hoststack tests and trending - all live now in csit.fd.io, hover info available now, and labels on y-axis improved 15:23:36 <mackonstan> #info Tibor - same changes applied to iterative pages and graphs 15:24:18 <mackonstan> #info Tibor, Dave - quic tests fixed, so all hoststack tests are now running, and displayed in csit.fd.io 15:25:45 <mackonstan> #info Peter, Tibor - mlx5/cx5 dpdk tests also running in trending + report, and displayed in csit.fd.io 15:27:54 <mackonstan> #info Kai - wireguard 10k tunnels - conversation with vpp contributor(s) continues - 37764: wireguard: under-load state determination update | https://gerrit.fd.io/r/c/vpp/+/37764 15:28:12 <mackonstan> #link https://gerrit.fd.io/r/c/vpp/+/37764 15:29:42 <mackonstan> #info Kai - continuing working on it, driving agreement, but won't be able to merge in time for v23.02, but will go into master and once there, Yulong will push a csit patch for test coveragfe 15:31:12 <mackonstan> #info Kai - Yulong has currently a private patch covering that 15:32:44 <mackonstan> #info Tibor, Peter - telemetry working now, so far no issues (planning memory optimizations for backend, frontend pretty much done for now) 15:35:19 <mackonstan> #info Kai - re gtpu, wireguard job_specs updates for v23.02 - will check with Yulong and send update over email 15:35:35 <mackonstan> #topic Physical Infrastructure 15:36:04 <mackonstan> #info Peter - decommissioning process started - skx perf testbeds and nomad machines - multi phase 15:37:24 <mackonstan> #info Peter - executing on the plan, details are in vexxhost support tickets, links below 15:37:31 <mackonstan> #info FD.io - Infrastructure refresh, part1 15:37:38 <mackonstan> #link https://support.vexxhost.com/hc/en-us/requests/363450 15:37:46 <mackonstan> #info FD.io - Infrastructure refresh, part2 15:37:53 <mackonstan> #link https://support.vexxhost.com/hc/en-us/requests/363464 15:39:09 <mackonstan> #info mackonstan - discussing with Nvidia team re them providing 200ge and 100ge cables with integrated pluggables 15:40:17 <mackonstan> #topic Trending and Test Failures 15:41:00 <mackonstan> #info Vratko - two new issues per wiki page: 3.1.1 csit-dpdk-perf-mrr-weekly-master-3n-snr fails due to a missing symlink, 3.1.2 wrong MAC address on lf_2n_clx_testbed27.yaml 15:41:09 <mackonstan> #link https://wiki.fd.io/view/CSIT/TestFailuresTracking 15:43:34 <mackonstan> #info Vratko - CSIT code better handling MTU is not going to make it to CSIT-2302, test failures and performance impact, so deferred to next release cycle 15:44:17 <mackonstan> #info Vratko - fix for this should be able to make it into the csit-2302 rls - 3.1.3 (M) 3n-alt: Tests failing until 40Ge Interface comes up 15:44:57 <mackonstan> #topic AOB 15:45:18 <mackonstan> #info Kai - Yulong just pushed two patches 15:45:30 <mackonstan> #info Kai - 38137: Fix issue in csit hoststack test | https://gerrit.fd.io/r/c/csit/+/38137 15:47:06 <mackonstan> #info KAi - to trim reviewers' list to Dave and Peter 15:47:28 <mackonstan> #info Kai - 37860: Add 1M flows test suites of VPP IP4 and VPP IP6 | https://gerrit.fd.io/r/c/csit/+/37860 15:47:37 <mackonstan> #info KAi - to trim reviewers' list to Dave and Peter 15:50:16 <mackonstan> #info Juraj - re new mlx NICs - need to confirm which models are getting allocated - consultation over email - clear preference for cx6, as no cx7 available in arm labs, so if any issues, no way to fix them 15:50:18 <mackonstan> #endmeeting