#fdio-csit: FD.io CSIT project meeting
Meeting started by mackonstan at 15:00:48 UTC
(full logs).
Meeting summary
- Agenda bashing (mackonstan, 15:00:53)
- Jan Gelety (jgelety,
15:00:58)
- Tibor Frank (tifrank,
15:03:08)
- Physical labs, testbeds (mackonstan, 15:04:28)
- Vratko Polak. (vrpolak,
15:04:33)
- no updates, all green. (mackonstan,
15:04:34)
- Input from projects - VPP (mackonstan, 15:04:53)
- Dave Wallace (dwallacelf,
15:05:50)
- re CSIT-2001 - 2n-skx, 3n-skx - issue with NDR
throughput still persists, feedback from VPP call is to publish PDR
and MRR numbers, and update rls notes about poor NDR numbers until
fixed. (mackonstan,
15:08:33)
- agreement in VPP call to aim for end of March
for v19.08.2 maintenance release. (mackonstan,
15:09:17)
- Input from projects - TSc (mackonstan, 15:09:36)
- Scapy/GPL issue is still in resolution phase -
RayK is coordinating agreement from copyright owners/file
authors (mackonstan,
15:10:37)
- CNCF CNF Testbed (mackonstan, 15:10:54)
- no update (mackonstan,
15:10:57)
- Maciek - will ask Michael Pederson from Intel
to join one of the future CSIT calls to provide update. (mackonstan,
15:11:35)
- Releases - CSIT-2001 (mackonstan, 15:11:52)
- Testing - Jan - apart from Skx - everything is
finished. (mackonstan,
15:12:38)
- Report - Tibor - apart from Skx - everything
has been added to report. (mackonstan,
15:13:07)
- Report - Dave - there is one patch outstanding
for hoststack, that needs to be added to report. (mackonstan,
15:13:28)
- Releases - CSIT-2001 - 2n-skx, 3n-skx (mackonstan, 15:14:31)
- https://gerrit.fd.io/r/c/csit/+/25724
(dwallacelf,
15:14:36)
- Skx testing - Jan - performance jobs running on
2n-skx and about to start running on 3n-skx. Will have >5 runs
per testbed type before next week Wednesday. (mackonstan,
15:16:38)
- Skx report - Tibor - exclusion of NDR graphs
and analysis already prepared, so will be included in next week
maintenance report. As the test data arrives it will be included in
draft version at master link. (mackonstan,
15:18:38)
- Skx report - Maciek - will update vpp and dpdk
rls notes regarding the Skx environment/ucode issues (10% PDR
regression, poor NDR throughput). (mackonstan,
15:19:32)
- Releases - CSIT-1908 maintenance (mackonstan, 15:19:55)
- vpp device tests running. (mackonstan,
15:20:34)
- vpp performance tests - requires environment
updates - Peter - looking to have it done this week unless
interruption, Jan willing to help. (mackonstan,
15:21:44)
- weekly test job - Jan - working on spec listing
all required tests (mackonstan,
15:22:32)
- Skx NDR t/shooting (mackonstan, 15:22:42)
- Intel has access to FD.io labs and executing
additional telemetry to track down the NDR issue on 2n-skx
tb23 (mackonstan,
15:23:37)
- Releases - CSIT-2005 (mackonstan, 15:25:04)
- dates aligned with VPP (mackonstan,
15:25:14)
- https://wiki.fd.io/view/CSIT/csit2005_plan
(mackonstan,
15:25:44)
- We should our best to do all environment
updates early in the cycle (mackonstan,
15:28:29)
- More vulnerabilities discovered impacting
processor ucode, we need to track ucode developments with respective
vendors. (mackonstan,
15:31:02)
- Add standing agenda topic for CSIT call to
check re changes in spectre-meltdown-checker.sh and processor ucode
updates. (mackonstan,
15:32:06)
- Govind - Taishan upgrade to -72 kernel with new
ucode load failed, need to check logs, reverted to -54 (mackonstan,
15:33:40)
- Operations (mackonstan, 15:33:54)
- job retries - stopped now, ci-mgmt patch from
Ed/Dave merged (mackonstan,
15:35:59)
- VPP code performance (mackonstan, 15:36:40)
- vhost tests failing, vpp-in-vm not starting,
Peter to check (mackonstan,
15:41:09)
- ipsec tnlhw failing, Jan + Neale tracked down
the offending patch (mackonstan,
15:41:51)
- Vratko - no real regression noticed in
trending (mackonstan,
15:42:05)
Meeting ended at 15:42:09 UTC
(full logs).
Action items
- (none)
People present (lines said)
- mackonstan (40)
- collabot` (3)
- dwallacelf (2)
- vrpolak (1)
- jgelety (1)
- tifrank (1)
Generated by MeetBot 0.1.4.