15:04:46 <mackonstan> #startmeeting 161214 csit weekly
15:04:46 <collabot> Meeting started Wed Dec 14 15:04:46 2016 UTC.  The chair is mackonstan. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:04:46 <collabot> Useful Commands: #action #agreed #help #info #idea #link #topic.
15:04:46 <collabot> The meeting name has been set to '161214_csit_weekly'
15:04:52 <mackonstan> #chair mackonstan
15:04:52 <collabot> Current chairs: mackonstan
15:04:53 <edwarnicke> mackonstan: Sending embassies to neighboring projects CSITs (ODL, OpNFV)
15:04:54 <jgelety1> #info: Jan Gelety
15:04:56 <tfherbert> #info Tom Herbert
15:05:12 <mklotton1> #info Matej Klotton
15:05:18 <sgmonroy> #info Sergio Gonzalez Monroy
15:05:33 <edwarnicke> #info edwarnicke
15:05:36 <pmikus> #info Peter Mikus
15:05:47 <tifrank> #info Tibor Frank
15:05:52 <selias> #info Samuel Elias
15:07:37 <dwallacelf> #info Dave Wallace
15:08:36 <mackonstan> #topic Inputs from FD.io projects
15:08:40 <Patrick___> #info Patrick Lu
15:09:01 <mackonstan> #topic 	•	VPP - any questions or clarifications for csit-dev?
15:10:10 <mackonstan> #info dwallacelf - maintenance of Vagrant as a working and easy-to-use VPP run & test environment - question was raised about flaky status thereof
15:10:26 <mackonstan> #info existing wiki docs are wrong, misleading
15:11:34 <mackonstan> #info consensus on vpp call that dwallacelf will put together a proposal for a standard way to build either a Vagrant box, or qcow image, or some other image of sorts
15:13:57 <mackonstan> #info same applies to csit 3-node environment
15:14:52 <mackonstan> #topic HoneyComb
15:15:48 <mackonstan> #info selias - producing HC TOC for csit1701 report. will put on wiki.
15:15:56 <mackonstan> #info TLDK
15:16:04 <mackonstan> #topic TLDK
15:16:34 <mackonstan> #info pmikus - reviews are still in progress
15:17:01 <mackonstan> #info pmikus - completed review and +1
15:17:25 <pmikus> #info TLDK patch for bootstrap: https://gerrit.fd.io/r/#/c/3959/
15:17:40 <mackonstan> #topic TSC - Sending embassies to neighboring projects CSITs (ODL, OpNFV) - Ed
15:17:51 <pmikus> #info TLDK patch for tests and libraries: https://gerrit.fd.io/r/#/c/3958/
15:18:14 <mackonstan> #info edwarnicke - TSC is asking for a more vigorous collaboration between related networking projects
15:18:43 <mackonstan> #info edwarnicke - get the conversation going
15:18:53 <mackonstan> #info csit volunteers needed
15:19:17 <mackonstan> #info edwarnicke will facilitate - first between TPLs
15:20:02 <mackonstan> #topic FD.io CSIT operational status
15:21:51 <mackonstan> #topic Test environment infra topics
15:22:03 <mackonstan> #topic vhost NDR results still low per csit1609 report note - CPU jitter fixes with CFS settings - status
15:24:15 <mackonstan> #info pmikus - most of code in csit now, one open bug
15:24:36 <mackonstan> #info improvements already seen in csit-vpp-verify perf trending plots
15:24:56 <mackonstan> #link https://jenkins.fd.io/view/csit/job/csit-vpp-verify-perf-master-nightly-all/plot/RFC2544%3ANDR/
15:25:08 <mackonstan> #topic Adding Centos - func VIRL
15:29:39 <mackonstan> #info tfherbert - significant progress, almost there, LOS < xmas
15:31:24 <mackonstan> #topic Performance tests
15:31:38 <mackonstan> #topic DPDK perf baseline pre VPP integration
15:32:20 <mackonstan> #info pmikus - all comments closed - pmikus and mmiklus gave +1s
15:32:40 <mackonstan> #info few more open comments that need to be addressed
15:33:04 <mackonstan> #topic trending baseline tests - one-core, multi-core
15:34:02 <mackonstan> #info tifrank - sent email with analysis, awaiting comments from the csit-dev list
15:34:38 <mackonstan> #topic Functional tests - vpp make test framework
15:34:48 <mackonstan> #info no updates
15:34:57 <mackonstan> #topic Use case tests
15:35:13 <mackonstan> #topic vhost use cases - see thread on csit-dev
15:36:28 <mackonstan> #info two TWS calls done, all minuted on ircbot
15:36:38 <mackonstan> #topic Functional tests - csit virl
15:36:44 <mackonstan> #info no additional updates
15:37:13 <mackonstan> #topic FD.io CSIT test automation for IXIA Traffic Generator - Patrick Lu - 10min
15:41:39 <mackonstan> #info Patrick___ doing a demo of developed .py tools using ixia python apis to find NDR/PDR of VPP in different interface and thread setups
15:41:48 <mackonstan> #info all driven from interactive CLI
15:41:56 <mackonstan> #info iow not touching IXIA GUI :)
15:42:48 <mackonstan> #info all environment setup is in .yaml file(s)
15:44:04 <mackonstan> #info pmikus - q: are you still depending on ixia python api library ?
15:44:19 <mackonstan> #info Patrick___ - yes.
15:46:02 <mackonstan> #info pmikus - q: config done via json yaml file. possible to do it via cli ?
15:46:19 <mackonstan> #info Patrick___ - possible, partially.
15:47:47 <mackonstan> #info mackonstan: are you okay to push this code into csit repo.
15:47:55 <mackonstan> #info Patrick___ - yes.
15:48:19 <mackonstan> #topic Performance tests - adding cryptodev tests - test scenarios - Sergio, Radu - 10min
15:48:27 <mackonstan> #info Sergio - providing update
15:49:35 <mackonstan> #info Kiriil Rybalchenko - providing update
15:53:23 <tfherbert> #info I have one more Centos question. Centos just released 7.3 in the middle of my VIRL image building. Should I release a Centos 7.3 image as well as the Centos 7.2 image?
15:54:14 <mackonstan> #info Kirill confirmed that ipv4 crypto-dev csit performance code is almost ready, will push patch soon, before next csit call - at least that's the plan.
15:54:26 <tfherbert> #info I will bring this to email list....
15:54:46 <mackonstan> #info sgmonroy - issues with csit rf performance or time consumption for large scale e.g. 1k tunnels
15:57:23 <mackonstan> #info per action from 2016-11-30 - following the initial baseline ip4 crypto-dev, then we need to agree what to test next
15:57:36 <mackonstan> #info will allocate 10..15min on next call to cover this.
15:58:36 <pmikus1> #info tfherbert - i think as long as you can do _any_ CentOS build succesfully i do not expect much changes in upgrading to higher version from compatibility point of view. It might be just new packages and chaging some variables.
15:59:14 <mackonstan> #endmeeting