#opnfv-cperf: cperf

Meeting summary

1. 1. dfarrell07 has a response from Veryx folks about test tool in CI (dfarrell07, 19:18:54)
   2. They have a Dockerfile and some similar, but it's all on an FTP server (dfarrell07, 19:20:38)
   3. need to either get them working with upstream tooling or do it for them (dfarrell07, 19:20:54)
   4. Can at least download Dockerfile and push to Gerrit, but would be better for them to do it. Maybe can add them as git also-by or similar? (dfarrell07, 19:21:24)
   5. dfarrell07 replied to NIkos about test in CI questions, sorry for delay. Seems like they have some good scripting done, about ready to move into CI. (dfarrell07, 19:21:54)
   6. Nikos will need help from Jamo on getting access to PODs (dfarrell07, 19:22:06)
   7. Jamo says Nikos will need to ask Jack about getting VPN access (dfarrell07, 19:28:09)
   8. Luis, Jamo et al say that we really want Veryx to do more to get their own tooling working, we're happy to take their tool but we have other good options, we need them to do the integration work (dfarrell07, 19:29:28)
   9. Jamo has POD working well! w00t! (dfarrell07, 19:31:07)
   10. Apex is deploying, controller is running, ready to run tests on it (dfarrell07, 19:31:19)
   11. Jamo is now trying to get CBench running from test container (dfarrell07, 19:31:48)
   12. There are issues with ROBOT tests expecting tools system and system with ODL to be different boxes, some stuff to figure out around passing IPs around (dfarrell07, 19:33:12)
   13. Can now click a button in Jenkins and wipe all, lay down Apex (dfarrell07, 19:35:37)
   14. Need to update Nikos with new status of PODs, they are ready for tests (dfarrell07, 19:36:04)
   15. Luis is working on OpenFlow test stuff still (dfarrell07, 19:38:58)
   16. Luis has clustering, scale in cluster, and longevity tests up next (dfarrell07, 19:40:47)
   17. Al has been talking to ETSI NFV folks (in China?) about metrics, "total path instantiation time" (dfarrell07, 19:44:03)
   18. There have to be things at the networking layer to get connectivity between VNFs (dfarrell07, 19:44:35)
   19. Batch model, using NB to controller to SBs, isn't a problem. Can verify flows and see if in good shape (dfarrell07, 19:45:08)
   20. With learning model, when only lay rules as flows pass through switches, more complex. How to time that? What if there are both kinds? (dfarrell07, 19:45:45)
   21. This is the kind of test that fits perfectly in CPerf, end-to-end in real env is ideal, not running just against ODL, not a stand-alone tool. Can work on adding soon-ish, now that we have Apex (dfarrell07, 19:51:06)

Action items

1. (none)

People present (lines said)

1. dfarrell07 (23)
2. collabot (3)