yahoo-eng-team team mailing list archive

[Launchpad Bug Tracker <1742505@xxxxxxxxxxxxxxxxxx>]

This bug was fixed in the package openvswitch - 2.8.1-0ubuntu0.17.10.2

---------------
openvswitch (2.8.1-0ubuntu0.17.10.2) artful; urgency=medium

  * d/p/dpif-kernel-gre-mtu-workaround.patch,
    d/p/dpif-netlink-rtnl-Use-65000-instead-of-65535-as-tunnel-MTU.patch:
    Cherry pick in-flight fixes for workaround to correctly set MTU
    of GRE devices via netlink (LP: #1742505).

openvswitch (2.8.1-0ubuntu0.17.10.1) artful; urgency=medium

  * New upstream stable release (LP: #1724622).

 -- James Page <james.page@xxxxxxxxxx>  Sat, 20 Jan 2018 10:22:31 +0000

** Changed in: openvswitch (Ubuntu Artful)
       Status: Fix Committed => Fix Released

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to neutron.
https://bugs.launchpad.net/bugs/1742505

Title:
  gre_sys set to default 1472 when using path_mtu > 1500 with ovs 2.8.x

Status in Ubuntu Cloud Archive:
  Fix Committed
Status in Ubuntu Cloud Archive pike series:
  Fix Committed
Status in Ubuntu Cloud Archive queens series:
  Fix Committed
Status in neutron:
  Invalid
Status in linux package in Ubuntu:
  Confirmed
Status in openvswitch package in Ubuntu:
  Fix Released
Status in linux source package in Artful:
  Confirmed
Status in openvswitch source package in Artful:
  Fix Released
Status in linux source package in Bionic:
  Confirmed
Status in openvswitch source package in Bionic:
  Fix Released

Bug description:
  [Impact]
  OpenStack Clouds using GRE overlay tunnels with > 1500 MTU's will observe packet fragmentation/networking issues for traffic in overlay networks.

  [Test Case]
  Deploy OpenStack Pike (xenial + pike UCA or artful)
  Create tenant networks using GRE segmentation
  Boot instances
  Instance networking will be broken/slow

  gre_sys devices will be set to mtu=1472 on hypervisor hosts.

  [Regression Potential]
  Minimal; the fix to OVS works around an issue for GRE tunnel port setup via rtnetlink by performing a second request once the gre device is setup to set the MTU to a high value (65000).

  
  [Original Bug Report]
  Setup:
  Pike neutron 11.0.2-0ubuntu1.1~cloud0
  OVS 2.8.0
  Jumbo frames setttings per: https://docs.openstack.org/mitaka/networking-guide/config-mtu.html
  global_physnet_mtu = 9000
  path_mtu = 9000

  Symptoms:
  gre_sys MTU is 1472
  Instances with MTUs > 1500 fail to communicate across GRE

  Temporary Workaround:
  ifconfig gre_sys MTU 9000
  Note: When ovs rebuilds tunnels, such as on a restart, gre_sys MTU is set back to default 1472.

  Note: downgrading from OVS 2.8.0 to 2.6.1 resolves the issue.

  Previous behavior:
  With Ocata or Pike and OVS 2.6.x
  gre_sys MTU defaults to 65490
  It remains at 65490 through restarts.

  This may be related to some combination of the following changes in OVS which seem to imply MTUs must be set in the ovs database for tunnel interfaces and patches:
  https://github.com/openvswitch/ovs/commit/8c319e8b73032e06c7dd1832b3b31f8a1189dcd1
  https://github.com/openvswitch/ovs/commit/3a414a0a4f1901ba015ec80b917b9fb206f3c74f
  https://github.com/openvswitch/ovs/blob/6355db7f447c8e83efbd4971cca9265f5e0c8531/datapath/vport-internal_dev.c#L186

To manage notifications about this bug go to:
https://bugs.launchpad.net/cloud-archive/+bug/1742505/+subscriptions