#fdio-csit: CSIT-weekly-160907
Meeting started by mackonstan at 14:04:21 UTC
(full logs).
Meeting summary
- agenda bashing (mackonstan, 14:04:49)
- Inputs from FD.io projects (mackonstan, 14:06:05)
- VPP update from yday call - next vpp release -
Keith Burns proposal for Damjan Marion to drive the release
(mackonstan,
14:07:02)
- DPDK+VPP focused pre-integration testing - Waterman Cao. (mackonstan, 14:08:00)
- open action pending - Waterman proposal: start
with opt-1 as step1 for Intel team to get familiar with CSIT code
and VPP environment. (mackonstan,
14:08:21)
- ACTION: Waterman to
review this high-level plan with Ray Kinsella, and come back with
the strawman plan (mackonstan,
14:08:32)
- TLDK test cases - Anna (Qun Wan). (mackonstan, 14:08:55)
- review done by mmiklus, Anna posted patches
based on feedback (mackonstan,
14:09:49)
- updated patch in review (mackonstan,
14:10:06)
- TBD: agreee on jenkins job creation/triggers
for tldk test cases (mmiklus,
14:11:23)
- FD.io CSIT operational status (mackonstan, 14:13:18)
- jenkins slaves capacity issues are gone for
now (mackonstan,
14:13:47)
- some issues starting vpp1206 images in perf
testbed (mackonstan,
14:14:39)
- this issue reported by Peter Mikus (mackonstan,
14:15:38)
- https://jira.fd.io/browse/CSIT-400
(mackonstan,
14:15:41)
- link to monitor fd.io jenkins slaves
status (mackonstan,
14:16:50)
- https://jenkins.fd.io/computer/
(mackonstan,
14:16:54)
- Build Queue should be empty or small
(mackonstan,
14:17:46)
- CSIT exec time reduction (mackonstan, 14:19:11)
- from jgelety - no progress on splitting the TCs
into separate jobs, will work on it in coming week (mackonstan,
14:19:32)
- csit1609 release readiness (mackonstan, 14:19:52)
- • CSIT 16.09 jenkins jobs (mackonstan, 14:22:02)
- perf numbers into report from
csit-vpp-perf-1609-all (mackonstan,
14:23:34)
- to generate the graphs, run
csit-vpp-perf-1609-all manually multiple times (mackonstan,
14:25:52)
- ACTION: Peter to run
csit-vpp-perf-1609-all manually, report issues, auto-produce very
nice graphs :) (mackonstan,
14:26:51)
- Status - https://jira.fd.io/issues/?filter=10667. (mackonstan, 14:31:03)
- Status - CSIT Jira 1609-p0. (mackonstan, 14:31:25)
- https://jira.fd.io/issues/?filter=10667
(mackonstan,
14:32:07)
- candidates for cherry-picking for csit1609:
csit-384, 383, 27 (mackonstan,
14:33:41)
- Status - performance tests (mackonstan, 14:34:56)
- comparing NDR, PDR results of vpp1609 vs.
vpp1606 (mackonstan,
14:36:39)
- ACTION: Maciek, Miro
- comparing NDR, PDR results of vpp1609 vs. vpp1606, aim to identify
code changes responsible (mackonstan,
14:38:30)
- VM-virtio, VPP-vhost-user numbers. (mackonstan, 14:40:16)
- current NDR, PDR results much lower than
expected, and observed in non-CSIT testbeds (mackonstan,
14:40:51)
- ACTION: Peter, Maciek
working on resolution to identify root cause and fix perf issues in
csit testbeds for vpp vhost-user with guest-VM running
test-pmd (mackonstan,
14:41:35)
- Latency measurements. (mackonstan, 14:42:01)
- latency results require further scrutiny and
analysis, as using brand new trex functionality with separate
latency measurements streams (mackonstan,
14:43:16)
- ACTION: Miro, Maciek,
dwallacelf - work with trex dev team (Yaroslav?) to drive
conclusions on latency measurements by trex in csit (mackonstan,
14:44:55)
- Review failing tests CSIT Jira 1609-fail (mackonstan, 14:46:41)
- csit1609 report will list failed test cases in
a separate section with explanations (mackonstan,
14:55:21)
- explanation will be pulled from csit JIRA for
tickets marked with 1609-fail label (mackonstan,
14:55:46)
- https://jira.fd.io/issues/?filter=10667&jql=project%20%3D%20CSIT%20AND%20resolution%20%3D%20Unresolved%20AND%20labels%20%3D%201609-fail%20ORDER%20BY%20priority%20DESC%2C%20updated%20DESC
(mackonstan,
14:56:03)
- • Tickets Blocked, Stuck In Review state, Red Flag Tickets, Yellow Flag Tickets. - redundant, already discussed (mackonstan, 14:57:05)
- Test report generation (mackonstan, 14:59:08)
- Peter updated parser, fixed text
rendering (mackonstan,
14:59:24)
- need to add performance graphs (mackonstan,
14:59:32)
- CSIT dev work plan - wiki plan, jira tracking (mackonstan, 15:00:08)
- next week fd.io TWS call will discuss vpp1612
release plan - Tue 13-Sep, 09:15 (mackonstan,
15:01:04)
- vpp1609 test gap analysis (mackonstan, 15:02:03)
- vpp test gap analysis (mackonstan, 15:02:52)
- followup required with Erin once back from
PTO (mackonstan,
15:03:06)
Meeting ended at 15:03:35 UTC
(full logs).
Action items
- Waterman to review this high-level plan with Ray Kinsella, and come back with the strawman plan
- Peter to run csit-vpp-perf-1609-all manually, report issues, auto-produce very nice graphs :)
- Maciek, Miro - comparing NDR, PDR results of vpp1609 vs. vpp1606, aim to identify code changes responsible
- Peter, Maciek working on resolution to identify root cause and fix perf issues in csit testbeds for vpp vhost-user with guest-VM running test-pmd
- Miro, Maciek, dwallacelf - work with trex dev team (Yaroslav?) to drive conclusions on latency measurements by trex in csit
People present (lines said)
- mackonstan (55)
- collabot (5)
- mmiklus (1)
Generated by MeetBot 0.1.4.