#fdio-meeting: FD.io CSIT Project Meeting
Meeting started by mackonstan at 14:04:17 UTC
(full logs).
Meeting summary
- Agenda bashing (mackonstan, 14:04:32)
- Vratko Polak. (vrpolak,
14:04:47)
- Maciek - new HW list proposed for FD.io lab
builds in 2024 (mackonstan,
14:05:00)
- Tibor Frank (tifrank,
14:05:05)
- Maciek Vratko - mlrsearch ietf bmwg draft-05
published last Monday (mackonstan,
14:06:18)
- Inputs from projects - VPP (mackonstan, 14:06:46)
- Andrew - vpp v23.10 release is today
(mackonstan,
14:07:03)
- Andrew - asking if any last minute patches into
vpp stable/2310 (mackonstan,
14:08:50)
- Andrew - going ahead with pressing the button
to merge rls notes and artefacts built (few hrs) and then into
package cloud (mackonstan,
14:09:30)
- Inputs from projects - TSC (mackonstan, 14:12:24)
- Vratko - nothing in the notes. (mackonstan,
14:12:40)
- E810 flow "offload" tests (mackonstan, 14:13:24)
- Kai - has caught up on the history of this
work (mackonstan,
14:13:37)
- Kai - sent email yesterday to vpp-dev with
question for Damjan, subject "help need for patch review and DPDK
flags support" (mackonstan,
14:15:17)
- HW accelerator - SPR QAT2.0 (mackonstan, 14:15:38)
- Dave Wallace (dwallacelf,
14:16:26)
- Kai - asking Peter if any chance to test the
patch. Peter was away, will look at this at the nearest
opportunity. (mackonstan,
14:16:44)
- HW accelerator - SNR QAT2.0 (mackonstan, 14:16:57)
- Kai - 2 systems to be shipped to Fd.io labs,
John DiGiglio scheduled for 24-25-Oct. It should be a replacement,
not a new ship, to avoid unnecessary custom costs. (mackonstan,
14:19:40)
- HW accelerator - SPR DSA (mackonstan, 14:21:59)
- Maciek - latest update from Intel/Georgii -
received a copy of BIOS v1.4 from SuperMicro but through a private
Intel-SuperMicro channel, Georgii clarifying with SuperMicro
when/where a public released version of BIOS supporting SPR DSA
accelerators will be released (or can be found - SuperMicro webpages
search does not yield any finds). (mackonstan,
14:24:17)
- HW accelerator - E810 GTPU offload (mackonstan, 14:24:31)
- Kai - this tests should be working, as no known
issues are known. (mackonstan,
14:25:01)
- Vratko Maciek - checked the csit.fd.io and the
gtpu-hw tests are actually running, see rc2 results in the link
below (mackonstan,
14:35:59)
- https://csit.fd.io/report/#eNrtVctuwjAQ_Jr0Um1lO5j00kMh_4GSePNQQ9h6HRA98O01CGmD-jhVRSq92JZnRzvr0cgcNh5XjP1TYhdJtkhM1rm4JOnzfdx8zybVCrZEYNKHePKVOSholLbFvM5qfSi1hXQogcmDVqpBQxoftapewZF7gY5mqzAOA_YM81kJugIMbbxtAo3t7oiXBSMMzh_7muW57wcRgroxCBqlXSBb9AJ-rVk41O6F8M0kwig8FkKZDCglAXmi8bOBpbb2xRq5e0MhxIcSvIoWCaSryy5hTxP0_IpZfqq4pp_07-cP-EnX9ZNvLZ_8t_PJt5ZP_rV82vxu2Pj16R-1-TuZhgZW
(mackonstan,
14:36:03)
- Kai - will review and analyse the results with
the Intel team, and confirm if they are matching expectations
(mackonstan,
14:37:00)
- Maciek - for now we will keep this item as an
open issue, until Kai confirms gtpu-hw results are as
expected (mackonstan,
14:37:40)
- CSIT-2310 release (mackonstan, 14:38:42)
- looking at rls2310 rc2 results spreadsheet
produced by Tibor (mackonstan,
14:41:55)
- Maciek - many af_xdp tests are failing with two
errors: Minimal rate loss ratio X does not reach target 0.0, T-Rex
STL runtime error. (mackonstan,
14:42:38)
- Vratko - tracking those, add to rls notes,
older jira ticket already there but it was rare, now no longer rare,
ticket should be updated. (mackonstan,
14:43:15)
- Vratko - 3n-alt tests - tests failing all or
high frequency with VPP crashing (no core files): SPD_FLOW_CACHE,
ipsec tun, gtpu. (mackonstan,
14:45:06)
- Vratko - 3n-icx - frequently failing:
reassembly (fragmentation issue with vpp fix not merged), wireguard
(expected) - both are known issues with tickets. (mackonstan,
14:47:24)
- Vratko - 3n-snr - async ipsec tests failing
(expected due to broken qat ipsec hw) (mackonstan,
14:49:48)
- Vratko - 3na-spr - wireguard (known)
(mackonstan,
14:50:52)
- Vratko - 3nb-spr - async (known, HW QAT
issue) (mackonstan,
14:51:13)
- Dave Vratko - hoststack - 3na-spr - TCP tests
failing (VPP not starting but only for TCP test - other tests work
fine) (mackonstan,
14:54:48)
- Dave Vratko - hoststack - 3nb-spr - interfaces
not coming up (mackonstan,
14:55:28)
- Physical Infra (mackonstan, 14:56:53)
- Maciek - Arm proposing Nvidia Grace-Grace
systems, Intel EMR replacement for SPR CPUs, possibly new E-Core
systems (mackonstan,
14:59:15)
- Trending and Test Failures (mackonstan, 14:59:50)
- Vratko - trending failures are matching what's
reported in 23.10 testing (mackonstan,
15:00:10)
Meeting ended at 15:01:26 UTC
(full logs).
Action items
- (none)
People present (lines said)
- mackonstan (42)
- collab-meetbot (9)
- dwallacelf (1)
- vrpolak (1)
- tifrank (1)
- jlinkes (0)
Generated by MeetBot 0.1.4.