yahoo-eng-team team mailing list archive

[Ihar Hrachyshka <1592017@xxxxxxxxxxxxxxxxxx>]

** Changed in: neutron
       Status: In Progress => Opinion

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to neutron.
https://bugs.launchpad.net/bugs/1592017

Title:
  ML2 uses global MTU for encapsulation calculation

Status in neutron:
  Opinion

Bug description:
  My goal is to achieve an MTU of 1500 for both Vlan and Vxlan based
  tenant networks.

  So I set the path_mtu to 1550, while leaving global_physnet_mtu at the
  default value of 1500.

  However, the MTU for my Vxlan networks is still calculated as 1450,
  because the underlying MTU is calculated as min(1500,1550)=1500 and
  then the encapsulation overhead is subtracted from that.

  IMHO the correct calculation would be to subtract the encapsulation
  overhead only from the path_mtu if it is specified and possibly take
  the minimum of that and global_physnet_mtu.

  If I set global_physnet_mtu to 1550, my Vxlan networks will be fine,
  but then the Vlan networks will get MTU 1550, which I also want to
  avoid.

To manage notifications about this bug go to:
https://bugs.launchpad.net/neutron/+bug/1592017/+subscriptions