15:02:14 <mackonstan> #startmeeting FD.io CSIT Project Meeting 15:02:14 <collab-meetbot> Meeting started Wed Mar 2 15:02:14 2022 UTC and is due to finish in 60 minutes. The chair is mackonstan. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:02:14 <collab-meetbot> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:02:14 <collab-meetbot> The meeting name has been set to 'fd_io_csit_project_meeting' 15:02:42 <mackonstan> #topic Agenda bashing 15:02:46 <tifrank> #info Tibor Frank 15:03:31 <dwallacelf> #info Dave Wallace 15:04:51 <mackonstan> #info Maciek Konstantynowicz 15:05:00 <mackonstan> #chair 15:05:00 <collab-meetbot> Current chairs: mackonstan 15:05:14 <mackonstan> #chair tifrank dwallacelf mackonstan 15:05:14 <collab-meetbot> Warning: Nick not in channel: tifrank 15:05:14 <collab-meetbot> Warning: Nick not in channel: dwallacelf 15:05:14 <collab-meetbot> Warning: Nick not in channel: mackonstan 15:05:14 <collab-meetbot> Current chairs: dwallacelf mackonstan tifrank 15:05:44 <pmikus> #info Peter Mikus 15:05:49 <mackonstan> #topic VPP update 15:06:01 <mackonstan> #info no updates 15:06:16 <mackonstan> #topic VSAP 15:07:06 <mackonstan> #info Dave, update from Florin, VSAP tests are working fine, but the VSAP dev team got reduced, and currently working on DMA/DSA for hoststack 15:07:14 <mackonstan> #topic TSC 15:07:46 <mackonstan> #info Maciek, got auth for buying few connectx5 NICs, vexxhost ticket open 15:08:04 <mackonstan> #info Dave, progressing purchase of DRAM for arm build machines 15:08:29 <mackonstan> #info Maciek, TSC chair voting is in progress 15:09:21 <mackonstan> #info Dave, make test refactoring in progress, addressing scapy gpl licensing 15:09:48 <mackonstan> #topic E810 flow "offload" tests 15:09:58 <mackonstan> #info no update 15:10:41 <mackonstan> #topic VPP v22.02 rc1 rc2 hot patch(es) items 15:11:20 <mackonstan> #info Maciek MTU handling and impact on tests tracked in csit and vpp tickets 15:12:18 <mackonstan> #info Maciek DPDK v21.11 introduced perf regression across the board, per discussions last week in vpp and csit calls, i think CSIT ticket tracking for that too 15:12:31 <fan_zhang> #info Fan Zhang 15:13:46 <mackonstan> #info Fan, no bandwidth / capacity to test proposed dpdk src workaround by dpdk folks 15:15:03 <mackonstan> #info Fan, also see email thread on dpdk-dev mailer 15:15:07 <fan_zhang> http://mails.dpdk.org/archives/dev/2022-March/236896.html 15:15:19 <fan_zhang> The link to DPDK mtu discussion therad 15:16:49 <mackonstan> #info Fan, will keep csit project posted on the fix progress 15:18:05 <mackonstan> #info Maciek, will refer to dpdk thread in csit jira ticket trackign the issue 15:18:25 <mackonstan> #topic csit-2202 15:19:26 <mackonstan> #info Tibor, all iterative jobs finished/ing, enough data to generate report 15:19:37 <mackonstan> #info Tibor, coverage tests started 15:20:46 <mackonstan> #info Tibor, draft report contains vpp v22.02 iterative data, and half-half of v22.02 - rc2 coverage data, per breakdown in email 15:20:58 <mackonstan> #link https://lists.fd.io/g/csit-dev/message/4465 15:21:43 <mackonstan> #info Tibor, report static conent, only rls notes to be updated 15:22:51 <mackonstan> #info Tibor, need to start to validate formatting/rendering of graphs/tables, counting on Viliam's, Maciek's, et al 15:30:08 <mackonstan> #info Viliam, processed all failures, same failures as in trending, 7/10 are intermittent, but 2 are failing new - i) 3n-icx ndr imix ipsec cbc can't find NDR, ii) af_xdp can't find NDR. 15:30:40 <mackonstan> #info Maciek re regressions, waiting for Vratko (away today) 15:32:03 <mackonstan> #info Tibor, predicting majority of coverage tests to be finished by eow, but single testbed configs will have a long tail of tests to execute 15:33:42 <mackonstan> #topic Physical Infrastructure 15:35:04 <mackonstan> #info Peter - yul1 DC received two of supermicro icx systems, two ampere ultra servers expected this friday 15:40:20 <mackonstan> #info Maciek, Peter - will send email re the next stage of perf systems build in yul1, two options: either 2n-icx perf testbed, or 3n-amp and 3n-snowridge 15:42:24 <mackonstan> #topic Virtual Infrastructure, CI/CD 15:44:20 <mackonstan> #info Dave, working with hICN to fix vpp dependencies for them 15:45:18 <mackonstan> #info Dave, Peter, nomad jenkins workers/slaves getting killed due to out of memory, intermittent .. 15:46:04 <mackonstan> #info Peter, observing ECC errors on arm machines 15:46:28 <mackonstan> #topic Environment, Framework 15:46:38 <mackonstan> #topic Trending 15:51:21 <mackonstan> #info Viliam, see failures reported in email to csit-report 15:51:24 <mackonstan> #link https://lists.fd.io/g/csit-report/message/5442 15:51:27 <mackonstan> #endmeeting