#fdio-meeting: FD.io CSIT project meeting
Meeting started by mackonstan at 14:59:40 UTC
(full logs).
Meeting summary
-
- mackonstan (mackonstan,
14:59:45)
- Tibor Frank (tifrank,
15:00:05)
- Vratko Polak. (vrpolak,
15:00:27)
- Peter Mikus (pmikus,
15:00:53)
- Viliam Luc (viliamluc,
15:01:04)
- Agenda bashing (mackonstan, 15:02:46)
- Juraj Linkeš (jlinkes,
15:02:59)
- Inputs from projects - VPP (mackonstan, 15:03:03)
- Andrew, need for v21.10.1, with one commit,
missing some header files needed for hicn, purely packaging, no csit
testing / involvement needed, planned for new Wed. (mackonstan,
15:04:18)
- Inputs from projects - VSAP (mackonstan, 15:04:30)
- no update (mackonstan,
15:04:36)
- Inputs from projects - TSC (mackonstan, 15:04:43)
- New or refactored tests (mackonstan, 15:05:23)
- E810 flow "offload" tests - Fan took action to
follow up - no concrete progress (mackonstan,
15:05:56)
- Scale tests: ip4, ip6, nat, ipsec. - Vratko -
WIP (mackonstan,
15:08:15)
- Dave Wallace (dwallacelf,
15:08:46)
- trending test duration - see this link
(mackonstan,
15:10:03)
- https://s3-docs.fd.io/csit/master/trending/stats/vpp_mrr.html
(mackonstan,
15:10:05)
- Releases - CSIT-2110 (mackonstan, 15:10:22)
- Tibor, all iterative runs done, coverage almost
(tx2, zen), report being generated - ETA 2.5hrs from now ca. 17:45
UTC (mackonstan,
15:11:51)
- https://s3-docs.fd.io/csit/rls2110/report/index.html
(mackonstan,
15:12:47)
- next week maintenance to include: i) all
coverage test data, ii) rca's by Vratko, iii) speedup graphs for
dpdk, iv) 3n-icx and possible more iterations from 2n-icx (no memif,
no vm, nat44ed failures), v) 2c iterative for aws in thput graphs,
vi) min/avg/max latency for coverage tests in aws. (mackonstan,
15:17:39)
- maciek: question, for aws ena 50g tests, why
trex is reporting min/avg/max latency data, but hdrhistogram data
has the same string for all tests with zero values (mackonstan,
15:19:35)
- vratko: will have a look at this trex
issue (mackonstan,
15:21:23)
- vratko: will open csit ticket, add to known
issues (mackonstan,
15:22:29)
- all - agreed this was the smothest release
ever. mostly down to very stable vinfra (single DC YUL1) and
thorough rc1 and rc2 testing - WELL DONE CSIT TEAM!!! (mackonstan,
15:24:47)
- Physical Infrastructure (mackonstan, 15:25:32)
- Virtual Infrastructure, CI/CD (mackonstan, 15:26:42)
- dave, looking nice and healthy, doing some
cleanup stuff (mackonstan,
15:26:53)
- dave, one of the big items is nexus shutdown,
will put together a checklist (mackonstan,
15:28:07)
- dave, will track the checklist through an LF IT
ticket (mackonstan,
15:28:26)
- Environment, Framework (mackonstan, 15:30:37)
- vratko, re vpp api checking, sent email listing
csit used api's that are not production (mackonstan,
15:31:27)
- vratio, flow, nat, ipsec async crypto, af_xdp
api (mackonstan,
15:31:57)
- vratko, mohsin is producing plugins/code making
use of damjan's thread-txq infra, and it's breaking some paths e.g.
memif (mackonstan,
15:33:28)
- 32471: interface: add multi tx-queues support for new tx infra |
https://gerrit.fd.io/r/c/vpp/+/32471
(vrpolak,
15:38:08)
- Trending (mackonstan, 15:43:02)
- viliam, no new failures in trending, looking to
improve the way failures are summarised in weekly emails to
csit-report (mackonstan,
15:45:27)
- juraj, re VPP-1995 i40e driver issue, Ray
confirmed that this will be fixed in the i40e driver, no ETA known
yet (mackonstan,
15:46:03)
- Housekeeping (mackonstan, 15:46:38)
- dave, wants to work with juraj to update
ci-management to make arm jobs run faster, by increasing the number
of cpus per job\ (mackonstan,
15:47:21)
Meeting ended at 15:50:34 UTC
(full logs).
Action items
- (none)
People present (lines said)
- mackonstan (37)
- collab-meetbot` (4)
- vrpolak (2)
- tifrank (1)
- pmikus (1)
- viliamluc (1)
- jlinkes (1)
- dwallacelf (1)
Generated by MeetBot 0.1.4.