yahoo-eng-team team mailing list archive

[Launchpad Bug Tracker <1929832@xxxxxxxxxxxxxxxxxx>]

This bug was fixed in the package neutron - 2:16.3.2-0ubuntu2

---------------
neutron (2:16.3.2-0ubuntu2) focal; urgency=medium

  * d/p/updates-for-python3.8.patch: Cherry-picked from
    https://review.opendev.org/c/openstack/neutron/+/793417
    to ensure py38 keepalived-state-change cleanup (LP: #1929832).
  * d/p/d/p/0001-L3-Check-agent-gateway-port-robustly.patch: Dropped.
    Fixed in 16.3.2 stable point release.

neutron (2:16.3.2-0ubuntu1) focal; urgency=medium

  * New stable point release for OpenStack Ussuri (LP: #1927976).

 -- Corey Bryant <corey.bryant@xxxxxxxxxxxxx>  Thu, 27 May 2021 15:54:50
-0400

** Changed in: neutron (Ubuntu Focal)
       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/1929832

Title:
  stable/ussuri py38 support for keepalived-state-change monitor

Status in Ubuntu Cloud Archive:
  Invalid
Status in Ubuntu Cloud Archive ussuri series:
  Fix Committed
Status in neutron:
  In Progress
Status in neutron package in Ubuntu:
  Invalid
Status in neutron source package in Focal:
  Fix Released

Bug description:
  [Impact]
  Please see original bug description. Without this fix, the neutron-l3-agent is unable to teardown an HA router and leaves it partially configured on every node it was running on.

  [Test Plan]
  * deploy Openstack ussuri on Ubuntu Focal
  * enable L3 HA
  * create a router and vm on network attached to router
  * disable or delete the router and check for errors like the one below
  * ensure that the following line exists in /etc/neutron/rootwrap.d/l3.filters:

  kill_keepalived_monitor_py38: KillFilter, root, python3.8, -15, -9

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

  The victoria release of Openstack received patch [1] which allows the
  neutron-l3-agent to SIGKILL or SIGTERM the keepalived-state-change
  monitor when running under py38. This patch is needed in Ussuri for
  users running with py38 so we need to backport it.

  The consequence of not having this is that you get the following when
  you delete or disable a router:

  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent [req-8c69af29-8f9c-4721-9cba-81ff4e9be92c - 9320f5ac55a04fb280d9ceb0b1106a6e - - -] Error while deleting router ab63ccd8-1197-48d0-815e-31adc40e5193: neutron_lib.exceptions.ProcessExecutionError: Exit code: 99; Stdin: ; Stdout: ; Stderr: /usr/bin/neutron-rootwrap: Unauthorized command: kill -15 2516433 (no filter matched)
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent Traceback (most recent call last):
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent   File "/usr/lib/python3/dist-packages/neutron/agent/l3/agent.py", line 512, in _safe_router_removed
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent     self._router_removed(ri, router_id)
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent   File "/usr/lib/python3/dist-packages/neutron/agent/l3/agent.py", line 548, in _router_removed
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent     self.router_info[router_id] = ri
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent   File "/usr/lib/python3/dist-packages/oslo_utils/excutils.py", line 220, in __exit__
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent     self.force_reraise()
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent   File "/usr/lib/python3/dist-packages/oslo_utils/excutils.py", line 196, in force_reraise
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent     six.reraise(self.type_, self.value, self.tb)
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent   File "/usr/lib/python3/dist-packages/six.py", line 703, in reraise
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent     raise value
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent   File "/usr/lib/python3/dist-packages/neutron/agent/l3/agent.py", line 545, in _router_removed
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent     ri.delete()
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent   File "/usr/lib/python3/dist-packages/neutron/agent/l3/dvr_edge_router.py", line 236, in delete
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent     super(DvrEdgeRouter, self).delete()
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent   File "/usr/lib/python3/dist-packages/neutron/agent/l3/ha_router.py", line 492, in delete
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent     self.destroy_state_change_monitor(self.process_monitor)
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent   File "/usr/lib/python3/dist-packages/neutron/agent/l3/ha_router.py", line 438, in destroy_state_change_monitor
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent     pm.disable(sig=str(int(signal.SIGTERM)))
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent   File "/usr/lib/python3/dist-packages/neutron/agent/linux/external_process.py", line 113, in disable
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent     utils.execute(cmd, run_as_root=self.run_as_root)
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent   File "/usr/lib/python3/dist-packages/neutron/agent/linux/utils.py", line 147, in execute
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent     raise exceptions.ProcessExecutionError(msg,
  2021-05-26 02:11:44.653 3457514 ERROR neutron.agent.l3.agent neutron_lib.exceptions.ProcessExecutionError: Exit code: 99; Stdin: ; Stdout: ; Stderr: /usr/bin/neutron-rootwrap: Unauthorized command: kill -15 2516433 (no filter matched)

  Which results in the router being deleted from neutron but not the
  node. In my case i had both a qrouter and snat ns left with IPs still
  configured as well as my fip ip rule allocation still present in
  /var/lib/neutron/fip-priorities

  [1]
  https://github.com/openstack/neutron/commit/4fb505891ee32ae41247f1d7a48b7455b342840e

  [Regression Potential]
  This change is backported from the stable/victoria release to authorize cleaning up of keepalived-state-chane via rootwrap [1] when running under python3.8. Where things can go wrong with l3.filters would be in the form of filter mistakes that allow or disallow running the intended command. In this case the code is picked straight from what is in stable/victoria and above and has already been tested by Ed, so it appears to have very low regression potential.
  [1] https://wiki.openstack.org/wiki/Rootwrap

To manage notifications about this bug go to:
https://bugs.launchpad.net/cloud-archive/+bug/1929832/+subscriptions