14:00:38 #startmeeting FD.io CSIT project meeting 14:00:38 Meeting started Wed Jun 19 14:00:38 2019 UTC. The chair is mackonstan. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:38 Useful Commands: #action #agreed #help #info #idea #link #topic. 14:00:38 The meeting name has been set to 'fd_io_csit_project_meeting' 14:00:47 #topic Agenda bashing 14:01:09 #info Jan Gelety 14:01:31 #chair 14:01:31 Current chairs: mackonstan 14:02:07 #info Dave Wallace 14:02:23 #info Tibor Frank 14:03:43 #topic FD.io CSIT lab infrastructure 14:03:51 #info no rep from Vexxhost present 14:04:23 #info but then no outstanding issue 14:04:36 #topic Inputs from LFN and FD.io projects 14:04:46 #topic VPP 14:05:34 #info Dave: 19.08 schedule published, but no content published on the wiki. 14:07:30 #info Peter: VPP DPDK version was bumped to 19.05 on 30-May 14:07:47 #info need to bump DPDK ver for tests involving testpmd and l3fwd 14:08:42 #info VPP is following a strict commit message template, looks good 14:09:24 #topic HoneyComb 14:09:37 #info no updates 14:09:47 #topic TSC 14:09:58 #info no updates 14:10:11 #topic Operational status 14:11:08 #info Vanessa via email: Jenkins slowness issues 14:12:26 #info Tibor some jobs and tests are failing for CSIT patches tests are failing 14:12:53 #info Tibor: roughly 20% of cases are affected 14:15:29 #into Tom: Centos CI status - patches merged, but don't see any Centos tests running on vpp_device, likely need to add Jenkins executor to run them 14:19:58 #info Need to ask Ed to add Centos executors to Nomad orchestrator. Going forward need to make container infra code accessible. 14:22:15 #topic VPP code performance 14:24:46 #info Maciek: Vratko latest patch extended daily mrr test coverage, per tag list here 14:24:57 #link https://git.fd.io/csit/tree/resources/libraries/bash/function/mrr-daily.txt 14:26:47 #info Maciek: from daily csit-report email: 3n-skx, 27 tests failed - all vhost, needs to be checked 14:27:07 #topic Releases 14:29:31 #info CSIT-1904 test data collection finished, 7 weeks after the first release 14:31:12 #info Maciek: suggest for 1908 the focus is on doing only baseline driver paths for complete set of NICs 14:31:54 #info Maciek: test selection for report jobs would need to be done using RF tags as is the case today, just that both RF tags in test suites and filters would need to be updated 14:33:57 #info CSIT-1908 plan 14:34:11 #link https://wiki.fd.io/view/CSIT/csit1908_plan 14:34:17 #info Dave: add TCP tests 14:36:09 #info Juraj: aiming to have both Arm performance daily jobs and vpp_device tests/jobs running 14:36:29 #info: Peter: some progress on Atom-Denverton front 14:39:34 #topic Developments - Framework 14:40:05 #info Tibor: VAT to PAPI migration - big patches merges, IPutils, … 14:40:22 #info Tibor: continuing with classify 14:40:39 #info Jan: IPsecutils, expecting done next week 14:43:09 #info Peter: re L1 KWs alignment between -func/-dev and -perf - working on teardowns, then on setups 14:44:59 #info Dave: re VPP-API "Flag Day" gate-keeping, testing latest patch from Vratko 14:46:21 #info Dave: vpp api-gen introduced default values, but they're not included in CRC calcs in papi generated .json files, so this needs to be fixed on the vpp side 14:47:14 #info Dave: next thing would be add email support, then we could start dry-runs 14:48:21 #topic Test environments 14:49:26 #info vpp_device tests - failed machine replaced, working well 14:51:53 #info Peter: but tests are failing due to VPP IP library rewrites - APIs changed 14:51:55 #link https://gerrit.fd.io/r/c/12296/ 14:52:50 #topic Performance trending and analytics 14:53:06 #info no updates 14:53:10 #topic New tests 14:54:09 #info Peter: IPsec tests - added 5k, 10k - working; 20k tunnels code in review 14:55:09 #info Peter: running into issues with handling large VAT/PAPI history logs 14:56:21 #info Peter: Peter: it takes 1.5min to program 10k tunnels into VPP.. worried about PAPI performance 14:57:41 #info Need to get an update from Vratko+Ole about PAPI over remote socket performance 14:58:23 #topic TRex 14:58:41 #info Peter: TRex running on Python3, no issues. 15:00:00 #endmeeting