#fdio-csit: 23-Aug FD.io CSIT Project Meeting

Meeting summary

1. 1. Peter Mikus (pmikus1, 14:00:16)
   
   
2. : Agenda bashing (mackonstan1, 14:00:22)
   1. Dave Wallace (dwallacelf, 14:00:26)
   
   
3. Inputs from LF and FD.io projects (mackonstan1, 14:05:29)
   1. Billy McFall (BillyM, 14:07:02)
   2. VPP project: reporting higher than expected number of rechecks in gerrit patches - see if we can automate RCA reporting of "recheck" comments use by VPP contributors (mackonstan1, 14:08:55)
   
   
4. : operational status (mackonstan1, 14:12:57)
   1. : DaveW reporting - 6hrs outage in LF infra with Jenkins minions… (mackonstan1, 14:13:14)
   2. : DaveW reporting - failure happened last week (mackonstan1, 14:13:37)
   
   
5. rls1707 experience (mackonstan1, 14:15:29)
   
6. Reported VPP Performance drop from 17.04 to 17.07 for l2bd-2vhost-1vm (mackonstan1, 14:15:55)
   1. Karl reporting: looking at flow scaling - 1k to 1M flows (mackonstan1, 14:17:08)
   2. Karl reporting: spot checking VPP releases - 17.04 and 17.07 concurrently (mackonstan1, 14:17:47)
   3. Karl reporting: mac learning with l2 bridge (mackonstan1, 14:18:07)
   4. Karl reporting: l2 bridge with 17.04 performance in line with expectations compared to earlier l2xc tests done earlier (mackonstan1, 14:18:45)
   5. Karl reporting: but 17.07 performance with scale 10k MAC upwards showed much lower throughput (mackonstan1, 14:19:11)
   6. Karl reporting: two bridges, bi-dir tests (mackonstan1, 14:19:27)
   7. Karl reporting: 1M flows in each direction, each pkt traverses each bridge (mackonstan1, 14:19:42)
   8. Karl reporting: 17.04 l2fib count ~4M (mackonstan1, 14:20:14)
   9. Karl reporting: 17.07 "show l2fib: 1067053 l2fib entries with 1048576 learned (or non-static) entries" (mackonstan1, 14:21:29)
   10. Karl provided the link to TRex configuration: (mackonstan1, 14:27:37)
   11. https://github.com/atheurer/lua-trafficgen (mackonstan1, 14:27:40)
   12. https://github.com/atheurer/lua-trafficgen/blob/master/trex-txrx.py (mackonstan1, 14:28:11)
   13. Karl reporting: test environment pretty much aligned with 170214 report forwarded by Tom Herbert to csit-dev (mackonstan1, 14:29:50)
   14. CSIT code patches to address the test coverage escapes: (mackonstan1, 14:46:18)
   15. https://gerrit.fd.io/r/#/c/8145/ (mackonstan1, 14:46:21)
   16. followup on csit-dev mailer (mackonstan1, 14:46:29)
   
   
7. rls1710 plan (mackonstan1, 14:46:56)
   1. https://wiki.fd.io/view/CSIT/csit1710_plan (mackonstan1, 14:51:36)
   
   
8. Performance tests (mackonstan1, 14:51:56)
   1. SW cryptodev tests - status - check with Sergio; (mackonstan1, 14:52:24)
   2. l2fwd-cryptodev - code was not ready - check with Sergio; (mackonstan1, 14:52:48)
   
   
9. VPP functional tests - virl (mackonstan1, 14:53:09)
   1. CSIT VIRL vs. VPP make-test - functional test coverage delta => proposal to improve vpp make-test coverage, and deprecate equivalent CSIT VIRL test (mackonstan1, 14:54:47)
   2. DaveW - proposal - schedule a TWS call to have the delta reviewed and decide on the course of action (mackonstan1, 14:55:17)
   
   
10. Honeycomb tests (mackonstan1, 14:55:36)
    
11. Test environment updates (mackonstan1, 14:55:51)
    1. adding test equipment: Netgate - Tom report - no activity on netgate (mackonstan1, 14:56:08)
    2. NestedVM refresh - see email thread on csit-dev: "[csit-dev] Nested VM refresh" (mackonstan1, 14:56:54)
    3. VIRL platform - VIRL hosts upgrade: tb4-virl3 still to be upgraded (mackonstan1, 14:57:48)
    4. need to be scheduled (mackonstan1, 14:58:04)
    5. Centos7 VIRL tests - status - no activity in last week (Tom reporting) (mackonstan1, 14:58:36)
    
    
12. design optimizations (mackonstan1, 14:58:55)
    1. Tibor reporting - no news (mackonstan1, 14:59:22)
    2. https://jira.fd.io/browse/CSIT-738 (mackonstan1, 14:59:24)
    
    
13. Housekeeping (mackonstan1, 14:59:42)

Action items

1. (none)

People present (lines said)

1. mackonstan1 (46)
2. collabot (4)
3. BillyM (1)
4. pmikus1 (1)
5. dwallacelf (1)