#fdio-meeting: FD.io CSIT Project Meeting
Meeting started by mackonstan at 14:01:37 UTC
(full logs).
Meeting summary
-
- Tibor Frank (tifrank,
14:01:49)
- Vratko Polak. (vrpolak,
14:01:50)
- Juraj Linkeš (jlinkes,
14:01:52)
- Peter Mikus (pmikus,
14:01:53)
- Agenda bashing (mackonstan, 14:02:34)
- Inputs from projects: VPP (mackonstan, 14:03:32)
- Andrew: v21.10 rls merge job in progress, some
issues with packagecloud connectivity (mackonstan,
14:04:02)
- Inputs from projects: VSAP (mackonstan, 14:04:38)
- no update (mackonstan,
14:04:47)
- Inputs from projects: TSC (mackonstan, 14:04:54)
- Maciek: Lijian from Arm accepted by TSC as a
TSC member (mackonstan,
14:05:57)
- Dave: licensing is slowly progressing
(mackonstan,
14:06:48)
- New or refactored tests (mackonstan, 14:07:11)
- Dave Wallace (dwallacelf,
14:07:21)
- E810 flow "offload" tests - Peter, MAciek -
vpp_device tests were failing and got removed from vpp-csit-verify
jobs. Asking Fan to RCA it. (mackonstan,
14:10:09)
- IPsec SPD flow cache patches - Juraj - outbound
traffic with flow cache tests retested, passed, and merged
(mackonstan,
14:11:02)
- https://gerrit.fd.io/r/c/csit/+/34256
(jlinkes,
14:11:43)
- Juraj - this is the patch for the new tests,
but not adding to job_specs yet, will add once out of csit-2110 rls
cycle (mackonstan,
14:12:25)
- Scale tests - Vratko - only missing merge of
34131, to complete the job_spec cleanup from tests that are
inflating job execution times beyond the set limits (mackonstan,
14:14:45)
- Releases CSIT-2110 (mackonstan, 14:16:27)
- Tibor, Maciek - Reducing quantity of tests
executed for daily and weekly trending, and for report (mackonstan,
14:18:41)
- Juraj - from Arm's point of view things are
looking good (mackonstan,
14:18:59)
- Tibor - rc2 runs are done, output in master
report being generated right now, incl. latest results from
AWS (mackonstan,
14:21:34)
- https://s3-docs.fd.io/csit/master/report/
(mackonstan,
14:21:58)
- updated report at above link in a few
hrs (mackonstan,
14:22:14)
- Tibor - expecting rls data in csit-2110 link to
be available tue or wed next week (mackonstan,
14:22:54)
- Tibor, Vratko, Maciek - discussion about
replacing rc2 data with rls test data, agreed for this to happen
once enough rls test result samples are available per testbed type
(5 is agreed minimum to provide meaningful stdev) (mackonstan,
14:32:17)
- Physical Infrastructure (mackonstan, 14:33:10)
- PEter, Maciek - tb24 fixed with NIC replaced
after going bad with FW upgrade (mackonstan,
14:35:21)
- Juraj - Ampere servers, may still be 1 or 2
mths away (mackonstan,
14:35:38)
- Maciek - ICX supermicro shipment tracking for
mid november (mackonstan,
14:35:55)
- Maciek discussing guesstimates for FD.io lab
budget planning for 2022 (mackonstan,
14:38:20)
- https://wiki.fd.io/view/CSIT/csit_lab_2022_plan
(mackonstan,
14:38:32)
- re 1. AWS Performance Testing - covered outside
of lab2022 budget (mackonstan,
14:39:52)
- re Build machines - Faster Arm - Dave proposing
to look at AWS Arm instances for build jobs (mackonstan,
14:42:54)
- re 3. Performance testbeds: testbeds - SPR
build priced for 2-node testbeds (no 3-node testbeds); Snowridge;
Arm (mackonstan,
14:52:00)
- re 4. Replace software Router with a hardware
Router - ignoring for now (mackonstan,
14:52:45)
- re 5. Testing Data Analytics - adding
placeholder for min 2 SPR servers for data backend vs AWS
alternatives (mackonstan,
14:53:26)
- re 2. Build machines - Faster Arm - adding
placeholder for 2 x Ampere server for build vs AWS
alternatives (mackonstan,
14:53:59)
- re 4. add warning that we need network topology
confirmed by Vexxhost before the final decision about getting or not
HW router (mackonstan,
14:55:09)
- Virtual Infrastructure, CI/CD (mackonstan, 14:55:20)
- Dave - CI running fine, no issues
noticed (mackonstan,
14:55:30)
- Dave - needs heads-up from Tibor re report
jobs, in order to merge disruptive ci-mgmt patches to disable
nexus (mackonstan,
14:56:55)
- Dave - few more steps before nexus can be
"powered down" (mackonstan,
14:58:42)
- Trending/Failures, anomalies (mackonstan, 15:19:38)
- VPP-1995 i40e driver issue, vlan stripping for
VF not working - Juraj, Peter, Maciek - Damjan confirmed this is
i40e driver issue with specific api change, details to be added to
the jira ticket. Maciek will consult with Intel FD.io project leads
to see what's the most optimal way to resolve this issue. Affects
all Xeon and Arm vpp_device testbeds upgraded to the new dpdk i40e
driver. (mackonstan,
15:25:51)
Meeting ended at 15:25:54 UTC
(full logs).
Action items
- (none)
People present (lines said)
- mackonstan (43)
- collab-meetbot` (4)
- jlinkes (2)
- tifrank (1)
- vrpolak (1)
- pmikus (1)
- dwallacelf (1)
Generated by MeetBot 0.1.4.