#fdio-csit: FD.io CSIT project weekly meeting
Meeting started by mackonstan at 14:01:33 UTC
(full logs).
Meeting summary
-
- Tibor Frank (tifrank,
14:02:00)
- Jan Gelety (jgelety,
14:02:14)
- Dave Wallace (dwallacelf,
14:02:14)
- Tom Herbert (tfherbert,
14:02:20)
- Vratko Polak. (vrpolak,
14:02:55)
- Agenda bashing (mackonstan, 14:03:54)
- FD.io CSIT lab infrastructure - skipping (mackonstan, 14:04:13)
- Juraj Linkeš (jlinkes,
14:04:21)
- Maciek: final BOM for LFN FD.io 2019 lab
extension provided - will send summary to csit-dev mailer.
(mackonstan,
14:05:33)
- Inputs from LFN and FD.io projects (mackonstan, 14:06:16)
- Dave: added standing agenda item to go over any
upcoming VPP API changes (mackonstan,
14:06:37)
- Operational status (mackonstan, 14:08:19)
- Peter Mikus (pmikus,
14:08:40)
- Number of trending jobs keep failing due to FIB
API incompatibility after VPP API change (mackonstan,
14:10:47)
- Tibor: didn't see any issues since last week,
issue went away "silently" (mackonstan,
14:12:49)
- Ed: LFN IT team rebooted Jenkins to work around
the load/slowness related issue (mackonstan,
14:13:34)
- see email from Vanessa re additional loggign
and Jenkins maintenance on 02-July (mackonstan,
14:14:08)
- Centos CI: Tom reporting - sent email re Nomad
execution, waiting for someone (Ed?) to pick up, will resend
email (mackonstan,
14:15:07)
- Ed: should be easy to resolve (mackonstan,
14:16:39)
- VPP code performance (mackonstan, 14:16:40)
- Jan: fixing IP FIB API that should solve the
incompatibility with VPP after their change (mackonstan,
14:18:39)
- https://gerrit.fd.io/r/c/20278/
(mackonstan,
14:18:42)
- Jan: all tests done, ready for merge, need
review(s) (mackonstan,
14:19:40)
- Releases (mackonstan, 14:20:12)
- CSIT-1908 plan (mackonstan,
14:21:58)
- https://wiki.fd.io/view/CSIT/csit1908_plan
(mackonstan,
14:22:14)
- Developments (mackonstan, 14:23:04)
- Framework - VAT to PAPI migration (mackonstan, 14:24:12)
- L1 KW refactoring VAT to PAPI - Tibor: all what
planned for June is done and merged. Added three more. (mackonstan,
14:25:08)
- L1 KW refactoring VAT to PAPI - Jan: finished
IPUtil, going back to IPsec, then next SRv6 libraries. (mackonstan,
14:25:46)
- Jan: issues with PAPI for scale tests - lots of
entries in PAPI history, quick fix is to print just first two lines
and last two lines for all scale tests (for MAC, IPv4, IPv6);
(mackonstan,
14:30:12)
- Jan: separate issue is the actual logging of
cmds - several times - need fix. Proposal over email. (mackonstan,
14:34:22)
- Vratko: PAPI over remote socket improves PAPI
performance, but still not fast enough due to synchronous operation
1k commands/sec (mackonstan,
14:36:02)
- Vratko: solution is asynchronous operation,
this needs some thinking (mackonstan,
14:36:33)
- Vratko: PAPI over remote socket - conflict with
msg indeces; doesn't happen w/o remote socket (mackonstan,
14:39:21)
- Vratko: suspecting issue/bug in VPP
(mackonstan,
14:39:36)
- https://jenkins.fd.io/view/csit/job/csit-vpp-perf-mrr-daily-master-3n-skx/496/
showing info_level logs increased to 1.13G for 227TCs (pmikus,
14:39:38)
- Framework - L1 KWs alignment between dev and perf (mackonstan, 14:40:15)
- Peter: progressing well for setup and
teardown (mackonstan,
14:40:31)
- Peter: some tests working, some more cleanup
required (mackonstan,
14:40:45)
- VPP-API "Flag Day" - CSIT-VPP gate-keeping (mackonstan, 14:41:19)
- Dave: Vratko is refactoring original
implementation to adapt to remote socket (mackonstan,
14:42:08)
- Alignment of DEV/PERF was split to 3 phases. 1)
Teardown -done, 2) Setup phase -done, 3) Configuration -in progress.
3rd part should be final for the ready-to-go solution (pmikus,
14:42:37)
- Discussion: how to integrate this into
vpp-csit- jobs - new job to be run first and stop other jobs,
similar to checkstyle jobs (mackonstan,
14:49:23)
- Test environments (mackonstan, 14:49:57)
- Performance trending and analytics (mackonstan, 14:50:41)
- New tests (mackonstan, 14:50:46)
- Vratko: 9000B tests for AVF got added via suite
generator change, based on VPP doc change (mackonstan,
14:51:51)
- TRex (mackonstan, 14:52:48)
- Latency HdrHistogram - code aligned with latest
TRex master, but no target rls confirmed yet (mackonstan,
14:53:19)
- Swati: joined Cisco recently, part of FD.io
release engineering team (mackonstan,
14:54:19)
Meeting ended at 14:57:07 UTC
(full logs).
Action items
- (none)
People present (lines said)
- mackonstan (45)
- collabot` (4)
- pmikus (3)
- vrpolak (1)
- jlinkes (1)
- jgelety (1)
- dwallacelf (1)
- tfherbert (1)
- tifrank (1)
Generated by MeetBot 0.1.4.