#fdio-csit: CSIT Project Meeting
Meeting started by mackonstan at 14:02:11 UTC
(full logs).
Meeting summary
-
- Dave Wallace (dwallacelf,
14:03:36)
- Peter Mikus (pmikus,
14:05:26)
- Tibor Frank (tifrank,
14:06:44)
- Agenda bashing (mackonstan, 14:06:57)
- Physical labs, testbeds (mackonstan, 14:07:48)
- SSD fail - Peter - one SSD RMAed, failed SUT
recovered running with one SSD. (mackonstan,
14:08:33)
- Peter - guesstimate is that 10-12 SSDs failed
to date, out of skx and clx, it is suspicious. Will compile the
list, we can then share it with LFN/Vexxhost for verification and
spares mgmt, as well as punt it to vendor/Intel for RCAing.
(mackonstan,
14:13:46)
- Inputs from projects - VPP (mackonstan, 14:14:05)
- Inputs from projects - VSAP (mackonstan, 14:14:35)
- Inputs from projects - TSC (mackonstan, 14:14:40)
- Inputs from projects - CNCF CNF Testbed (mackonstan, 14:15:00)
- Were mainly focusing on control/orchestration
layer, not much on performance, still using VPP, Michael Pedersen
will contact us once any assistance from FD.io required (mackonstan,
14:16:12)
- Releases - CSIT-2001 (mackonstan, 14:16:32)
- 2n-skx - Jan - 8/10 perf runs done, data
available for graphs and compare tables. Full tests to be
done. (mackonstan,
14:17:42)
- 3n-skx - Jan - 1/10 perf run done, next ones
are running, delayed by testbed failures. (mackonstan,
14:18:19)
- Jan - hoststack tests completed on
3n-hsw. (mackonstan,
14:18:34)
- Releases - CSIT-1908 maintenance (mackonstan, 14:19:06)
- Jan - jobs ready, MRR weekly run successfully,
only 9 tests failed, 6 l2patch, 3 xxx, need to investigate.
(mackonstan,
14:20:28)
- Jan - preference to wait with 1908.2 rls
testing until we are out of CSIT-2001 2n-skx, 3n-skx (mackonstan,
14:21:17)
- Releases - CSIT-2005 (mackonstan, 14:21:37)
- Peter - VPP native virtio tests are ready for
review (mackonstan,
14:22:33)
- Vratko - fast scale tests using async PAPI,
ready for review. e.g. ip4scale1m - VAT O(10sec), PAPI today
O(30min), asyncPAPI O(~1min). (mackonstan,
14:24:39)
- Operations (mackonstan, 14:25:42)
- vpp-csit-device tests - re-enable voting -
Vratko - need to upgrade the current process of handling APIs,
blocker for voting (mackonstan,
14:26:50)
- VPP code performance (mackonstan, 14:27:53)
- vhost tests - Peter - latest dpdk changed api,
got fixed on vpp side. (mackonstan,
14:29:27)
- avf interface init - Peter - two problems: i)
csit infra issue CSIT-xxxx, ii) jira VPP-1858 (mackonstan,
14:30:39)
- srv6 tests failing - Peter - to be
investigated (mackonstan,
14:31:11)
- srv6 regression - Vratko - vpp patch fixes
this (mackonstan,
14:32:18)
- two new anomalies - Vratko - could be csit or
csit+vpp (mackonstan,
14:32:38)
- 2n-clx cxx556a - Vratko - TRex was taking
longer to send packets vs. config, aka tx-duration-stretching, fixed
by reducing Tx rate and adding TRex cores. Previous trending numbers
were incorrect. Effects MRR tests. (mackonstan,
14:34:34)
- Developments (mackonstan, 14:34:52)
- Jan - API functional test coverage - focusing
on IPsec tests, then SRv6 (mackonstan,
14:37:54)
Meeting ended at 14:39:48 UTC
(full logs).
Action items
- (none)
People present (lines said)
- mackonstan (33)
- collabot` (4)
- pmikus (1)
- jgelety (1)
- dwallacelf (1)
- tifrank (1)
Generated by MeetBot 0.1.4.