#fdio-csit: FD.io CSIT project meeting
Meeting started by pmikus at 13:58:20 UTC
(full logs).
Meeting summary
- Peaople presence (pmikus, 13:58:48)
- Peaople presence (pmikus, 13:59:03)
- Peter Mikus (pmikus,
13:59:36)
- Tibor Frank (tifrank,
13:59:37)
- Juraj Linkeš (jlinkes,
14:01:06)
- Agenda bashing (pmikus, 14:01:22)
- no inputs (pmikus,
14:02:37)
- Inputs from LFN and FD.io projects (pmikus, 14:02:53)
- DMM new patch , review in process (pmikus,
14:03:55)
- https://gerrit.fd.io/r/#/c/15189/ (pmikus,
14:04:41)
- CSIT operational status (pmikus, 14:05:11)
- VPP integration issue: server looses
connectivity when frequently installing and uninstalling during
single job. Investigation in progress. (pmikus,
14:07:42)
- VPP code performance (pmikus, 14:08:37)
- mackonstan (mackonstan,
14:10:01)
- alerting issue - Tibor update - attachment
issue has been solved (mackonstan,
14:15:11)
- Tibor - alert emails generated after each
trending jobs - one per testbed type, twice a day => 6 emails a
day (mackonstan,
14:16:18)
- correction - avg 6 emails every 24hrs
(mackonstan,
14:16:46)
- sent to csit-report fd.io mailer (mackonstan,
14:17:02)
- no updates (mackonstan,
14:18:29)
- https://wiki.fd.io/view/CSIT/csit1810_plan
(mackonstan,
14:18:54)
- branch pulled by Jan (mackonstan,
14:19:39)
- https://git.fd.io/csit/log/?h=rls1810
(mackonstan,
14:19:56)
- rc2 date today (mackonstan,
14:21:49)
- Soak Tests Experimental - Vratko - still WIP
and in gerrit draft. (mackonstan,
14:24:28)
- VPP_Device - Peter - successful job runs in LF
FD.io sandbox - were successful. (mackonstan,
14:24:59)
- some issues with porting this to production
Jenkins - Ed Kern. (mackonstan,
14:25:51)
- VPP_Device - Peter - initial implementation
still WIP and in gerrit, but not draft. Few comments to be
addressed, merge ETA before release close, hopefully. (mackonstan,
14:27:47)
- VPP_Path - is Open, but the concept of runing
VIRL test cases on the new VPP_Device infra has been proven.
Expecting to port most (or even all) VIRL tests, note however the
first set will include 2-Node tests only. (mackonstan,
14:29:42)
- Expecting to port most (or even all) VIRL
tests, note however the first set will include 2-Node tests only, in
next release i.e. CSIT-1901. (mackonstan,
14:30:23)
- Per VPP Patch Performance Checks - Vratko -
implemented next set of improvements, but his issue with VPP
reinstall. If lucky it should be finished by official rls date (i.e.
today +7days). (mackonstan,
14:31:18)
- CSIT PAPI Support - Peter on behalf on JanG -
most of the work is good and done. Peter and Ole Troan are doing the
PAPI keywords review. Remote execution of PAPI API calls (over REST
or other) is the next feature request, not supported today in VPP.
Will be part of PAPIv2. (mackonstan,
14:32:55)
- Graphs Layout Improvements - Tibor - moving to
final state. Information clarity improvements for throughput,
latency and speedup multicore. Gerrit patch ready for final
review. (mackonstan,
14:34:25)
- https://gerrit.fd.io/r/15155
(tifrank,
14:34:34)
- mackonstan to provide proposed testcase
grouping per graph instance in graphs_improvements.rst or separate
file if many lines. (mackonstan,
14:37:57)
- Clock cycles per VPP node into CSIT-PAL -
mackonstan - no update. (mackonstan,
14:38:36)
- Arm testbeds - Juraj - no status change.
(mackonstan,
14:39:26)
- new TWS calls every Friday at the same time as
Thursday TWS call.. (mackonstan,
14:40:39)
- mackonstan to update csit TWS call page with
relevant details - calls are zoom (mackonstan,
14:41:04)
- https://wiki.fd.io/view/CSIT/TWS
(mackonstan,
14:41:20)
- Atom testbeds - Yulong - due to the correct
Denverton servers not arriving to LFN FD.io lab in time (mackonstan,
14:43:15)
- separate off-site 2-Node testbed is in
operation, it's running CSIT framework with T-Rex (mackonstan,
14:43:49)
- generating output.xml files can be provided for
inclusion into CSIT-18.10 report, with obvious disclaimer that all
tests were not conducted within LFN FD.io physical infrastructure,
but were done off-site at Intel premises using FD.io CSIT code as
stored within FD.io project. (mackonstan,
14:45:27)
- Yulong shared a screen with MW Word doc listing
the MRR and NDR throughput results. (mackonstan,
14:47:45)
- all results are based on VPP-18.10 rc2
(mackonstan,
14:49:01)
- Yulong will publish rc2 base test results data,
ETA <24-Oct, if no issues. (mackonstan,
14:50:33)
- Peter - errors shown by Yulong may be due to
old firmware in NIC cards, as similar was observed in CSIT
testbeds. (mackonstan,
14:51:24)
- Peter - ASTF interactive progressed quite
significantly, need to find time to analyse (mackonstan,
14:52:53)
- FD.io minii-summit before Kubecon US in
Seattle, it's on Monday 10-Dec (mackonstan,
14:54:36)
- Trex releases https://github.com/cisco-system-traffic-generator/trex-core/blob/master/doc/release_notes.asciidoc
(pmikus,
14:55:14)
- https://fd.io/event/fd-io-mini-summit-at-kubecon-cloudnativecon-seattle-wa/
(mackonstan,
14:55:35)
- https://docs.google.com/forms/d/e/1FAIpQLSf7q2XKYhsYYgqtCYOnisyu5RAfhQiKNHzrNugq0AqIXdHB-A/viewform
(mackonstan,
14:55:56)
- looks like CFP may be still open (mackonstan,
14:56:05)
- Tibor Frank (tifrank,
14:00:09)
- Juraj Linkeš (jlinkes,
14:00:47)
- Peter Mikus (pmikus,
14:02:00)
- Jan Gelety (jgelety,
14:03:50)
- VPP - released today, we can can start
testing (tifrank,
14:03:54)
- HoneyComb - func tests ready for 1810, no perf
no present (tifrank,
14:04:55)
- DMM - https://gerrit.fd.io/r/#/c/15189/ ready
for merge (tifrank,
14:07:08)
- LF infra - no issues (tifrank,
14:08:18)
- Performance trending - no issues (tifrank,
14:09:49)
- Soak Tests Experimental - ready (tifrank,
14:10:54)
- VPP_Device - code ready, execution in Jenkins
WiP (tifrank,
14:11:35)
- Per VPP Patch Performance Checks -
merged (tifrank,
14:11:59)
- CSIT PAPI Support - still WiP (tifrank,
14:13:02)
- Graphs Layout Improvements - WiP (tifrank,
14:14:04)
- ARM - testbeds ready, perf tests WiP
(tifrank,
14:16:39)
- ARM HW in lab, needs to wired (tifrank,
14:19:49)
- Running tests: It will take 15 days in the best
case, 19 days realistic (tifrank,
14:22:17)
Meeting ended at 14:23:28 UTC
(full logs).
Action items
- (none)
People present (lines said)
- mackonstan (49)
- tifrank (24)
- pmikus (15)
- collabot (5)
- jlinkes (2)
- jgelety (1)
Generated by MeetBot 0.1.4.