#opnfv-cperf: cperf

Meeting summary

1. 1. /me started taking notes late, sorry (dfarrell07, 18:33:57)
   2. Discussion about ongoing on-mail discussions about how to do some tests, about issues with L2switch bug/strange feature, about future ways of doing network-restart-type tests (dfarrell07, 18:34:41)
   3. Al suggests that instead of doing OF test that involves programming a lot of flows via a network restart (like we did in whitepaper), we could have the network already configured/connected and then cut a part of the network so the controller must program a lot of flows (dfarrell07, 18:35:41)
   4. We may be seeing an issue with mininet during network restart test. mininet is itself a bit slow, ODL can handle programming new switches as they come up, keeping up with mininet and not getting as stressed as we'd like (dfarrell07, 18:36:27)
   5. Discussion about stats collection in OFP (dfarrell07, 18:36:36)
   6. Luis has been helping ONOS folks with how we do stats collection (dfarrell07, 18:37:06)
   7. New OFP in ODL is doing stats quite well, see whitepaper number (dfarrell07, 18:37:38)
   8. https://lists.opendaylight.org/pipermail/integration-dev/2016-May/006623.html See this mail thread for info about Nikos' work/questions about L2Switch behavior (dfarrell07, 18:38:48)
   9. Jamo has been heads down in ODL Int/Test (and getting his first Java patch in, congrats!), no major update (dfarrell07, 18:39:24)
   10. dfarrell07 is in Spain teaching Ericsson folks about open source/testing, following/writing discussions and whitepaper/blog updates, trying to get back to NSTAT but not so much this week (dfarrell07, 18:40:27)
   11. Luis asked asked Brocade L2 switch ppl to prio helping Nikos with issue described in mail above (dfarrell07, 18:40:56)
   12. Quite a bit of discussion about realistic vs stressing tests, how to find balance (dfarrell07, 18:42:35)
   13. Network restart and moving many flows in a network are nice and realistic, good candidates for next whitepaper (2nd, we did 1st in 1.0 and 1.1) (dfarrell07, 18:43:57)

Action items

1. (none)

People present (lines said)

1. dfarrell07 (15)
2. collabot` (3)