Previously, Networking packaging excluded OCF resources for the 'neutron-netns-cleanup' and 'neutron-ova-cleanup' services.
As a result, Pacemaker did not clean up unused network namespaces and Open vSwitch resources when fencing the Networking service.
With this update, the previously excluded OCF resources were added to Networking packages.
As a result, Pacemaker is now able to clean up unused network namespaces and Open vSwitch resources when fencing the Networking service.
DescriptionMiguel Angel Ajo
2014-07-11 14:20:30 UTC
Description of problem:
Those services are one-shot, that are started before all the agents
and 'stopped' after all the agents.
start means: cleanup non used resources (ovs, netns)
stop means: cleanup all resources (ovs.netns) which is used when
moving agents to a different host.
Using only systemd, we could't get to work the stop semantics, because
systemd discovers there's no actual PID running, and won't accept the
stop semantics.
To test the resource definitions are installed, you may need to
install pacemaker and then run:
# pcs resource describe ocf:neutron:NetnsCleanup
ocf:neutron:NetnsCleanup - neutron netns cleanup resource agent
This resource agent does nothing during execution, only executes a cleanup during start, and a force cleanup during stop of the netns resources generated by neutron agents.
# pcs resource describe ocf:neutron:OVSCleanup
....
...
tested with openstack-neutron-2014.1.1-4.el7ost.noarch
[root@a25400702875 ~]# pcs resource describe ocf:neutron:NetnsCleanup
ocf:neutron:NetnsCleanup - neutron netns cleanup resource agent
This resource agent does nothing during execution, only executes a cleanup during start, and a force cleanup during stop of the netns resources generated by neutron agents.
[root@a25400702875 ~]# pcs resource describe ocf:neutron:OVSCleanup
ocf:neutron:OVSCleanup - neutron OVS cleanup resource agent
This resource agent does nothing during execution, only executes a cleanup during start, and a force cleanup during stop of the openvswitch resources generated by neutron agents.
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.
http://rhn.redhat.com/errata/RHBA-2014-0936.html
Description of problem: Those services are one-shot, that are started before all the agents and 'stopped' after all the agents. start means: cleanup non used resources (ovs, netns) stop means: cleanup all resources (ovs.netns) which is used when moving agents to a different host. Using only systemd, we could't get to work the stop semantics, because systemd discovers there's no actual PID running, and won't accept the stop semantics.