yahoo-eng-team team mailing list archive

[Tom Carroll <Thomas.Carroll@xxxxxxxx>]

Public bug reported:

Version: Kilo
Compute hypervisor: XenServer 6.5SP1
Compute vm: Ubuntu 14.04.3

This issue appears first in kilo--and not before--when running XenServer
6.5SP1 hypervisor. In this environment, root-helper is set to /usr/bin
/neutron-rootwrap-xen-dom0, which executes commands in the hypervisor's
Dom0 context. This problem keeps the neutron-openvswitch-agent from
verifying that the ovsdb-client is currently executing. Additionally,
the XMLRPC interface design for communicating to DOM0 does not work with
``streaming'' data pipelines.

The error message is reported every couple seconds:
2016-01-05 10:16:25.255 1728 ERROR neutron.agent.linux.ovsdb_monitor [req-8037e215-c9d3-4593-b0d8-a0fc1bba6c7d - - - - -] Interface monitor is not active
2016-01-05 10:16:27.295 1728 ERROR neutron.agent.linux.ovsdb_monitor [req-8037e215-c9d3-4593-b0d8-a0fc1bba6c7d - - - - -] Interface monitor is not active
2016-01-05 10:16:29.249 1728 ERROR neutron.agent.linux.ovsdb_monitor [req-8037e215-c9d3-4593-b0d8-a0fc1bba6c7d - - - - -] Interface monitor is not active

The cause of the problem is that AsyncProcess.pid is None and
AsyncProcess.is_active returns false when root_helper runs commands in
another system context.  Consequently, the command is respawned every
ovsdb_monitor_respawn_interval seconds.

The new behaviour began with commit
3543d8858691c1a709127e25fc0838e054bd34ef.

Furthermore, the introduced changes no longer works with the
communication pattern of neutron-rootwrap-xen-dom0. This is an XML rpc
style interfaces that supports request/response-style communication.

** Affects: neutron
     Importance: Undecided
         Status: New


** Tags: monitor neutron ovsdb xenserver

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to neutron.
https://bugs.launchpad.net/bugs/1531245

Title:
  neutron.agent.linux.ovsdb_monitor errors, agent fails to add 1Q tag to
  new interfaces

Status in neutron:
  New

Bug description:
  Version: Kilo
  Compute hypervisor: XenServer 6.5SP1
  Compute vm: Ubuntu 14.04.3

  This issue appears first in kilo--and not before--when running
  XenServer 6.5SP1 hypervisor. In this environment, root-helper is set
  to /usr/bin/neutron-rootwrap-xen-dom0, which executes commands in the
  hypervisor's Dom0 context. This problem keeps the neutron-openvswitch-
  agent from verifying that the ovsdb-client is currently executing.
  Additionally, the XMLRPC interface design for communicating to DOM0
  does not work with ``streaming'' data pipelines.

  The error message is reported every couple seconds:
  2016-01-05 10:16:25.255 1728 ERROR neutron.agent.linux.ovsdb_monitor [req-8037e215-c9d3-4593-b0d8-a0fc1bba6c7d - - - - -] Interface monitor is not active
  2016-01-05 10:16:27.295 1728 ERROR neutron.agent.linux.ovsdb_monitor [req-8037e215-c9d3-4593-b0d8-a0fc1bba6c7d - - - - -] Interface monitor is not active
  2016-01-05 10:16:29.249 1728 ERROR neutron.agent.linux.ovsdb_monitor [req-8037e215-c9d3-4593-b0d8-a0fc1bba6c7d - - - - -] Interface monitor is not active

  The cause of the problem is that AsyncProcess.pid is None and
  AsyncProcess.is_active returns false when root_helper runs commands in
  another system context.  Consequently, the command is respawned every
  ovsdb_monitor_respawn_interval seconds.

  The new behaviour began with commit
  3543d8858691c1a709127e25fc0838e054bd34ef.

  Furthermore, the introduced changes no longer works with the
  communication pattern of neutron-rootwrap-xen-dom0. This is an XML rpc
  style interfaces that supports request/response-style communication.

To manage notifications about this bug go to:
https://bugs.launchpad.net/neutron/+bug/1531245/+subscriptions