yahoo-eng-team team mailing list archive

[Slawek Kaplonski <1859362@xxxxxxxxxxxxxxxxxx>]

As there is no reply from the author of this RFE since some time and as we didn't really saw real use case for this when we last time discussed that, I will now mark this rfe as not accepted.
Feel free to reopen it to add to the drivers meeting agenda when You will provide use case(s) which we were asking for.

** Changed in: neutron
       Status: New => Incomplete

** Changed in: neutron
       Status: Incomplete => 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/1859362

Title:
  Neutron accepts arbitrary MTU values for networks

Status in neutron:
  Opinion

Bug description:
  Openstack allows the user to configure a different MTU per network,
  with the default taken from the ML2 neutron::plugins::ml2::path_mtu
  value ( minus VXLAN overhead, if used )

  (overcloud) [stack@vim1 ~]$ openstack network set --mtu 1500
  management

  However, currently (Queens at least) the user can specify arbitrary values
  (overcloud) [stack@vim1 ~]$ openstack network set --mtu 99999 management
  (overcloud) [stack@vim1 ~]$ openstack network set --mtu 10 management

  The maximum possible MTU is 64K, but most network equipment won't support more than 9200-9300.
  Even when using Jumbo frames and VXLAN encapsulation, practical deployments typically wouldn't use more than 9050 byte MTUs. RFC 791 requires a minimum MTU of 68, and IPv6 requires at least 1280

  Proposal: Limit the accepted values to 1280-9250

To manage notifications about this bug go to:
https://bugs.launchpad.net/neutron/+bug/1859362/+subscriptions