#fdio-csit: Weekly CSIT Project Meeting

Meeting summary

1. 1. mackonstan (mackonstan, 14:02:39)
   2. Peter Mikus (pmikus1, 14:02:41)
   3. Jan Gelety (jgelety1, 14:02:46)
   
   
2. Agenda bashing (mackonstan, 14:02:56)
   1. Marek Gradzki (mgradzki, 14:03:23)
   2. Juraj Linkes (jlinkes, 14:04:23)
   
   
3. VPP update (mackonstan, 14:05:41)
   1. api freeze on 9-Jul (mackonstan, 14:05:50)
   2. Damjan adding memory heap allocation with numa affinity to workers, expecting positive impact on performance (mackonstan, 14:09:51)
   
   
4. HC update (mackonstan, 14:09:56)
   1. Marek - API freeze in 2 weeks time, want to discuss impact of VIRL migration on HC tests. (mackonstan, 14:13:41)
   
   
5. CSIT operational status (mackonstan, 14:13:59)
   1. no issues reported (mackonstan, 14:14:04)
   
   
6. VPP code performance (mackonstan, 14:14:19)
   1. ipsec scale tests failing, ip scale tests failing (mackonstan, 14:16:13)
   
   
7. Releases (mackonstan, 14:18:17)
   
8. New integration tests (mackonstan, 14:22:58)
   1. VPP_Device - weekly TWS calls to continue (mackonstan, 14:23:19)
   2. VPP_Path: CSIT_VIRL migration - Jan - progressing okay, issues with vpp classify tests (test_classifier.py) (mackonstan, 14:26:24)
   3. https://docs.google.com/spreadsheets/d/1PciV8XN9v1qHbIRUpFJoqyES29_vik7lcFDl73G1usc/edit?usp=sharing (mackonstan, 14:26:36)
   4. Juraj - issues with testing tap in VPP make_test, not supported. Should consider using VPP_Device for tap driver tests. (mackonstan, 14:33:00)
   5. Juraj - looking into parallelising VPP_make_test execution. Collaborating with Klement. (mackonstan, 14:34:40)
   6. HC test topology (mgradzki, 14:35:03)
   7. https://docs.fd.io/csit/master/report/honeycomb_functional_tests/overview.html (mgradzki, 14:35:08)
   8. HC got tested in Containers, so not expecting any major issue (mackonstan, 14:39:10)
   9. consider using VPP_Device for HC functional tests (mackonstan, 14:39:46)
   10. more details to be covered on the next TWS VPP_Device call (mackonstan, 14:44:03)
   
   
9. VPP code performance monitoring (mackonstan, 14:44:34)
   1. Per VPP patch tests - Vratko - burst MRR tests are still being tuned to yield more precise detection (mackonstan, 14:45:32)
   2. PAL is already prepared for processing burst MRR results (mackonstan, 14:49:56)
   
   
10. VAT to PAPI (mackonstan, 14:51:38)
    1. for CSIT_VIRL this is already addressed by moving to VPP make test (mackonstan, 14:53:32)
    2. for VPP_Path_Device performance tests - would require refactoring L1 KWs (mackonstan, 14:54:23)
    
    
11. Tests (mackonstan, 14:54:31)
    1. MLRsearch - expecting to move to MLR in 18.07 (mackonstan, 14:55:17)
    2. Vratko - provided prototype MLR tests for Skylake 2-node testbeds (mackonstan, 14:56:02)
    
    
12. TRex (mackonstan, 14:56:19)
    1. Peter - hitting issues with xxv710 2p25GE, would need to use similar trick as with xl710 2p40ge. (mackonstan, 14:58:28)
    
    
13. New FD.io CSIT lab (mackonstan, 14:58:41)
    1. 4 * mcx556a-edat NICs just arrived (mackonstan, 14:59:59)
    2. Peter - issues with 3 servers => 3 tickets open (mackonstan, 15:00:20)
    3. failures affecting 3 testbeds :( (mackonstan, 15:00:41)
    4. Arm - Juraj - performance testbeds are finally coming together, hoping for 3-Nde perf testbed ready next week (mackonstan, 15:01:36)
    
    
14. DMM CSIT status (mackonstan, 15:03:42)
    1. Sharath - ubuntu1604 jobs work, centos7 jobs fail. (mackonstan, 15:04:07)

Action items

1. (none)

People present (lines said)

1. mackonstan (42)
2. collabot` (4)
3. mgradzki (3)
4. jgelety1 (1)
5. jlinkes (1)
6. pmikus1 (1)