I've just redeployed a cloud onto Xenial/Ocata in a similar configuration; the instances running on the cloud all get the higher MTU as configured by neutron, irrespective of the MTU on the tunnel bridge.
I've confirmed this using iperf to assess performance between units including mismatched MTU configurations/reality. In terms of charm configuration using: juju config neutron-api network-device-mtu=9000 juju config neutron-gateway instance-mtu=8500 should not be required any longer; if you just set global-physnet- mtu=9000 neutron will do the maths to figure out overlay overheads and drop MTU's for network segments as appropriate. Just as a side note; the versions referenced in the original bug report are Newton versions, not Ocata versions - maybe there is a different in function between the two releases. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/1674307 Title: Neutron v9 jumbo frames support To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/neutron/+bug/1674307/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
