Difference between revisions of "CSIT/Meeting"
From fd.io
< CSIT
(REVERT: Add link for AMD performance comparison via trending graphs.) |
(Correct AMD perf link.) |
||
| Line 52: | Line 52: | ||
#** Waiting for updated spectre-meltdown script: https://github.com/speed47/spectre-meltdown-checker/pull/418 | #** Waiting for updated spectre-meltdown script: https://github.com/speed47/spectre-meltdown-checker/pull/418 | ||
#* AMD performance results review - FD.io CSIT labs vs. AMD CSIT labs, Keesang et al | #* AMD performance results review - FD.io CSIT labs vs. AMD CSIT labs, Keesang et al | ||
| − | #** Current state in CSIT lab: | + | #** Current state in CSIT lab: http://csit.fd.io/trending/#eNrdkssKwjAQRb-m3ZSBZGrblQsf_yExHR_Q1pBEqX69qRSmXYgiIthNQjh37kwu4_zJ0sZRNY-yZVQsIyyOZTiidJGE62IMYAO6akEKsSc0UrdZliuwZa2gQshnW5AayB_Cq9Vb5QhqazsLXHUW5dmP_JiYw5XJ8y5coCwprqiQiSc36DIchjU7q2pyxxuxMAzPXIckGEk9dvdXM6D9D4v1Q_FhbDjB2PALsd0a_MG2Penyv9v2IjacYGzvblu2jp1X1s9RIIJAQEyECLL-iKkpe5aDTEfsDg9kWyM |
# Releases [https://wiki.fd.io/view/CSIT/csit2206_plan CSIT-2206] | # Releases [https://wiki.fd.io/view/CSIT/csit2206_plan CSIT-2206] | ||
#* rc1 testing status | #* rc1 testing status | ||
Revision as of 14:32, 15 June 2022
Meeting Details
Weekly on Wednesday 15:00 Greenwich, London time.
FD.io Zoom Meeting room (passcode <remove spaces> '4 1 2 3 2 4')
If no DST in Europe and US: 07:00 PST, 10:00 EST, 15:00 UTC, 15:00 UK/GMT, 16:00 CET, 17:00 Israel Std Time, 20:30 India Std Time, 23:00 China Std Time.
If DST in Europe: 07:00 PDT, 10:00 EDT, 14:00 UTC, 15:00 UK/BST, 16:00 CEST, 17:00 Israel Summer Time, 19:30 India Std Time, 22:00 China Std Time.
If no DST in Europe and DST in US: 08:00 PDT, 11:00 EDT, 15:00 UTC, 15:00 UK/GMT, 16:00 CET, 17:00 Israel Std Time, 20:30 India Std Time, 23:00 China Std Time.
Topic: FD.io CSIT Project Meeting
Join from PC, Mac, Linux, iOS or Android: https://zoom.us/my/fastdata?pwd=Z3Z0UnJyUmRIMlU3eTJLcGF6VEptQT09
Or iPhone one-tap (US Toll): +14086380968,,5301185804# or +16465588656,,5301185804#
Or Telephone:
Dial: +1 408 638 0968 (US Toll) or +1 646 558 8656 (US Toll)
+1 855 880 1246 (US Toll Free)
+1 877 369 0926 (US Toll Free)
Meeting ID: 530 118 5804
International numbers available: https://zoom.us/zoomconference?m=ppBOQMQTVxGYmbxNsVemC6KNo8eX2ptF
Issues connecting to the meeting? Contact: Maciek Konstantynowicz <mkonstan@cisco.com>.
- IRC
- #fdio-meeting room on libera
- More info: https://lists.fd.io/g/csit-dev/message/4345
- Web-based IRC client: https://web.libera.chat/
- #fdio-meeting room on libera
Agenda
The next CSIT meeting is scheduled for Wednesday 15-Jun 15:00-16:00 Greenwich, London time / 07:00-08:00 PST. Proposed topics:
- Agenda bashing.
- Inputs from projects
- FD.io: VPP, TSC, other.
- FD.io: **vppcfg** new project proposal - readout to CSIT postponed
- New or being refactored tests
- E810 flow "offload" tests
- device: updating tests with new dpdk offload flags, Fan.
- perf: work in progress, Fan.
- MERGED 36120: Add gtpu flow offload test suite | https://gerrit.fd.io/r/c/csit/+/36120
- OPEN 36119: Add ipsec policy test suites with flow director or rss accelerated | https://gerrit.fd.io/r/c/csit/+/36119
- Scale tests revisited
- Currently amount of executed scale tests is reduced for ip4, ip6, nat, ipsec.
- This is due to poor performance of PAPI and excessive time taken for configuring VPP.
- Work to restart to review feasible option to have this fixed, Vratko, Ole, Andrew, Dave.
- E810 flow "offload" tests
- VPP and CSIT hot items
- Spectre-BHB security vulnerability
- docs: Add references to Spectre-BHB security vulnerability
- Waiting for updated spectre-meltdown script: https://github.com/speed47/spectre-meltdown-checker/pull/418
- AMD performance results review - FD.io CSIT labs vs. AMD CSIT labs, Keesang et al
- Spectre-BHB security vulnerability
- Releases CSIT-2206
- rc1 testing status
- Report status
- CSIT and VPP 22.02 known issues status check, https://lists.fd.io/g/csit-dev/message/4516
- 100 GbE tests with newly fitted and calibrated MLX ConnectX5-ex and TRex v2.96
- with 16 pmd threads max aggregate 2x 100 GbE STL performance: 188Mpps TX rate and 133Mpps RX rate
- big thanks to Peter for the work with Hanoh to get it sorted finally
- telemetry - vpp perfmon and linux bpf telemetry
- CSIT UTI
- New platforms and testbeds: Ampere Altra 3n-alt, SnowRidge (3n-snr)
- MLRsearch tunings
- CVL 100 GbE flow-based tests
- Physical Infrastructure
- Performance testing: Xeon Skx/Ckx/Icx, Arm, Atom.
- Functional driver testing: Xeon Skx, Arm Tx2.
- Updated LFN FD.io infra support documentation.
- Lab new hardware builds:
- Perf - 17 supermicro icx servers
- 5 servers in production: 2n-icx, 3n-icx testbeds.
- 2 servers used for TG/TRex machines for new testbed builds (2n-alt).
- 1 server available for new build
- 9 servers finally arriving ...
- Perf - 2x Ampere Altra servers - onboarded
- Perf - 17 supermicro icx servers
- AMD requests to increase AMD participation level in CSIT benchmarking, Keesang et al.
- 2x Atom SnowRidge servers (4RU) shipped by Intel, in transit.
- Virtual Infrastructure, CI/CD
- LFN FD.io CI/CD systems.
- Jenkins jobs: csit-vpp, vpp-csit, trending.
- vpp-csit jobs: vpp-csit-verify-device-{..}-tx2 (not voting), vpp-csit-verify-device-{..}-skx (voting)
- Environment, Framework
- VPP API changes' tracking.
- VPP CLI usage in CSIT.
- Functional checks, e.g. "show nat44 summary" to count established sessions.
- VPP changed output in: 34877: nat: TCP state tracking based on RFC 7857/RFC 6146 | https://gerrit.fd.io/r/c/vpp/+/34877
- CSIT test were failing until fix: 35295: fix(nat): parse new show nat44 summary output | https://gerrit.fd.io/r/c/csit/+/35295
- Functional checks, e.g. "show nat44 summary" to count established sessions.
- Trending
- Hoststack tests: Requested tests to be added are the same as CSIT Report Hoststack Iterative Job Specs
- Daily perf tests: mrr-daily test selection, daily csit-report.
- Performance trending: Dashboard, Failures.
- Trendline graphs: L2, IP4, IP6, Vhost, L2 memif, Overlay, Crypto.
- New dynamic dashboard (CSIT UTI).
- Failures, anomalies, breakages
- port 0 is down when running traffic profile.
- all 1518B AVF crypto tests failed with no traffic.
- AF-XDP - NDR tests failing from time to time.
- VPP does not recognize X553/X557-AT2 10GE.
- multicore AVF tests are failing when trying to create interface.
- Not all DET44 sessions have been established: 4128767 != 4128768.
- VPP-1995 i40e driver issue, vlan stripping for VF not working.
- Analytics, Presentation.
- Fd.io Events.
- Housekeeping.
- Maintainers, commit message template, https://www.conventionalcommits.org
