yahoo-eng-team team mailing list archive

[Jake Yip <1751188@xxxxxxxxxxxxxxxxxx>]

Hi Slawek,

Thanks for the pointer! Had a brief look at master code and it seems to
do what we want. Unfortunately we are at Newton, but I'm glad that it's
fixed!

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

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to neutron.
https://bugs.launchpad.net/bugs/1751188

Title:
  neutron mtu setting in database for provider networks

Status in neutron:
  Invalid

Bug description:
  In our environment we have multiple sites, each with their own flat
  provider networks. We also have a centralised neutron-server, but each
  site runs their neutron-dhcp-agent service.

  I've been looking to upgrade some networks to MTU 9000, keeping others
  as is. What I want to do is for some neutron-dhcp-agents to dish out
  DHCP with MTU 9000 to the VMs.

  From the documentation (https://docs.openstack.org/mitaka/networking-guide/config-mtu.html), it seems like the right way to go is to set 
  `global_physnet_mtu = 9000` and, 

  for all other networks (except mtu 9000 one)
  `physical_network_mtus = network1:1500,network2:1500`

  If I am doing this right, I find the approach to be a bit problematic:

  1) It is inconsistent. Creation of network is via API but setting is
  in a config file. It would be nice to have API able to do both.

  2) If a site provider creates a new provider network, it will be
  defaulted to 9000 unless a corresponding change is made in the config
  file.

  I am wondering if there is any reason for the current behaviour, and
  if it is possible to have MTUs set via API / DB.

To manage notifications about this bug go to:
https://bugs.launchpad.net/neutron/+bug/1751188/+subscriptions