Comment 50 for bug 1734320

@Jeremy,

I just went through the entire series of comments in this bug. I also looked at the bug that Miguel Ajo filed a few days ago (https://bugs.launchpad.net/neutron/+bug/1767422) and the associated information in https://bugzilla.redhat.com/show_bug.cgi?id=1558336. My comments / questions are the following:

1) The work that Miguel Ajo is doing as a consequence of https://bugs.launchpad.net/neutron/+bug/1767422 has to do with the agents and performance issues. Although it is related to what we have been discussing in this bug, it doesn't address it

2) Reading carefully Sean's comment in #43 above, I notice that he states the following: "if you are currently able to transmit/recive packet in a vm that is attached to vlan 4095 i think you have a ovs bug. vlan 4095 is reserved as the null/dead vlan and ovs is drop all packet to/from interfaes with a tag of 4095...". So he seems to have written his comments (specifically the recommendation about ovs-ofctl mod-port down|up) under the assumption that one of the problems to be solved is that VMs can receive traffic even when their interfaces have been put on the dead vlan (4095). As far as I can tell, nobody has reported that here, which means that the patches proposed by Ihar in #37 and #38 are worth trying. Ihar also reports in #40 that the Neutron patch back-ports as is to Ocata and all the way to Liberty with some adjustments. So even though, according to Sean, there are use cases that might not be covered, the patches in #37 and #38 start us in the right direction mitigating the problem. Once they are in place, we can start nibbling at os-vif to work on the use cases that according to Sean are not fixed. I'd be happy to take a stab at that

3) Am I correct in my understanding that at this point we can submit the patches in #37 and #38 under the public / normal process in Gerrit? As soon a I get a confirmation of this, I will submit the patches for review