[Bug 1894843] Re: [dvr_snat] Router update deletes rfp interface from qrouter even when VM port is present on this host

[Launchpad Bug Tracker]

This bug was fixed in the package neutron -
2:17.1.0+git2021012815.0fb63f7297-0ubuntu3

---------------
neutron (2:17.1.0+git2021012815.0fb63f7297-0ubuntu3) hirsute; urgency=medium

  [ Edward Hope-Morley ]
  * Backport fix for dvr-snat missig rfp interfaces (LP: #1894843)
    - d/p/0001-Fix-deletion-of-rfp-interfaces-when-router-is-re-ena.patch

 -- Corey Bryant <corey.bryant at canonical.com>  Wed, 24 Feb 2021 12:41:20
-0500

** Changed in: neutron (Ubuntu Hirsute)
       Status: Triaged => Fix Released

-- 
You received this bug notification because you are a member of Ubuntu
OpenStack, which is subscribed to Ubuntu Cloud Archive.
https://bugs.launchpad.net/bugs/1894843

Title:
  [dvr_snat] Router update deletes rfp interface from qrouter even when
  VM port is present on this host

Status in Ubuntu Cloud Archive:
  Triaged
Status in Ubuntu Cloud Archive ussuri series:
  Triaged
Status in Ubuntu Cloud Archive victoria series:
  Triaged
Status in neutron:
  Fix Released
Status in neutron package in Ubuntu:
  Fix Released
Status in neutron source package in Focal:
  Triaged
Status in neutron source package in Groovy:
  Triaged
Status in neutron source package in Hirsute:
  Fix Released

Bug description:
  [Impact]
  When neutron schedules snat namespaces it sometimes deletes the rfp interface from qrouter namespaces which breaks external network (fip) connectivity. The fix prevents this from happening.

  [Test Case]
   * deploy Openstack (Ussuri or above) with dvr_snat enabled in compute hosts.
   * ensure min. 2 compute hosts
   * create one ext network and one private network
   * add private subnet to router and ext as gateway
   * check which compute has the snat ns (ip netns| grep snat)
   * create a vm on each compute host
   * check that qrouter ns on both computes has rfp interface
   * ip netns| grep qrouter; ip netns exec <ns> ip a s| grep rfp
   * disable and re-enable router
   * openstack router set --disable <router>;  openstack router set --enable <router>
   * check again
   * ip netns| grep qrouter; ip netns exec <ns> ip a s| grep rfp

  [Regression Potential]
  This patch is in fact restoring expected behaviour and is not expected to
  introduce any new regressions.

  -------------------------------------------------------------------------

  Hello,

  In the case of dvr_snat l3 agents are deployed on hypervisors there
  can be race condition. The agent creates snat namespaces on each
  scheduled host and removes them at second step. At this second step
  agent removes the rfp interface from qrouter even when there is VM
  with floating IP on the host.

  When VM is deployed at the time of second step we can lost external
  access to VMs floating IP. The issue can be reproduced by hand:

  1. Create tenant network and router with external gateway
  2. Create VM with floating ip
  3. Ensure that VM on the hypervisor without snat-* namespace
  4. Set the router to disabled state (openstack router set --disable <router>)
  5. Set the router to enabled state (openstack router set --enabled <router>)
  6. The external access to VMs FIP have lost because L3 agent creates the qrouter namespace without rfp interface.

  Environment:

  1. Neutron with ML2 OVS plugin.
  2. L3 agents in dvr_snat mode on each hypervisor
  3. openstack-neutron-common-15.1.1-0.20200611111910.7d97420.el8ost.noarch

To manage notifications about this bug go to:
https://bugs.launchpad.net/cloud-archive/+bug/1894843/+subscriptions