8.3. Configuring Fencing

Configuring fencing consists of (a) specifying one or more fence devices in a cluster and (b) specifying one or more fence methods for each node (using a fence device or fence devices specified).

Note

It is recommended that you configure multiple fencing mechanisms for each node. A fencing device can fail due to network split, a power outage, or a problem in the fencing device itself. Configuring multiple fencing mechanisms can reduce the likelihood that the failure of a fencing device will have fatal results.
Based on the type of fence devices and fence methods required for your configuration, configure cluster.conf as follows:
  1. In the fencedevices section, specify each fence device, using a fencedevice element and fence-device dependent attributes. Example 8.3, “APC Fence Device Added to cluster.conf ” shows an example of a configuration file with an APC fence device added to it.
  2. At the clusternodes section, within the fence element of each clusternode section, specify each fence method of the node. Specify the fence method name, using the method attribute, name. Specify the fence device for each fence method, using the device element and its attributes, name and fence-device-specific parameters. Example 8.4, “Fence Methods Added to cluster.conf ” shows an example of a fence method with one fence device for each node in the cluster.
  3. For non-power fence methods (that is, SAN/storage fencing), at the clusternodes section, add an unfence section. This ensures that a fenced node is not re-enabled until the node has been rebooted. When you configure a device that requires unfencing, the cluster must first be stopped and the full configuration including devices and unfencing must be added before the cluster is started. For more information about unfencing a node, see the fence_node(8) man page.
    The unfence section does not contain method sections like the fence section does. It contains device references directly, which mirror the corresponding device sections for fence, with the notable addition of the explicit action (action) of "on" or "enable". The same fencedevice is referenced by both fence and unfence device lines, and the same per-node arguments should be repeated.
    Specifying the action attribute as "on" or "enable" enables the node when rebooted. Example 8.4, “Fence Methods Added to cluster.conf ” and Example 8.5, “cluster.conf: Multiple Fence Methods per Node” include examples of the unfence elements and attributed.
    For more information about unfence see the fence_node man page.
  4. Update the config_version attribute by incrementing its value (for example, changing from config_version="2" to config_version="3">).
  5. Save /etc/cluster/cluster.conf.
  6. (Optional) Validate the updated file against the cluster schema (cluster.rng) by running the ccs_config_validate command. For example: 
    [root@example-01 ~]# ccs_config_validate 
Configuration validates
  7. Run the cman_tool version -r command to propagate the configuration to the rest of the cluster nodes. This will also run additional validation. It is necessary that ricci be running in each cluster node to be able to propagate updated cluster configuration information.
  8. Verify that the updated configuration file has been propagated.
  9. Proceed to Section 8.4, “Configuring Failover Domains”.
If required, you can configure complex configurations with multiple fence methods per node and with multiple fence devices per fence method. When specifying multiple fence methods per node, if fencing fails using the first method, fenced, the fence daemon, tries the next method, and continues to cycle through methods until one succeeds.
Sometimes, fencing a node requires disabling two I/O paths or two power ports. This is done by specifying two or more devices within a fence method. fenced runs the fence agent once for each fence-device line; all must succeed for fencing to be considered successful.
More complex configurations are shown in the section called “Fencing Configuration Examples”.
You can find more information about configuring specific fence devices from a fence-device agent man page (for example, the man page for fence_apc). In addition, you can get more information about fencing parameters from Appendix A, Fence Device Parameters, the fence agents in /usr/sbin/, the cluster schema at /usr/share/cluster/cluster.rng, and the annotated schema at /usr/share/doc/cman-X.Y.ZZ/cluster_conf.html (for example, /usr/share/doc/cman-3.0.12/cluster_conf.html).

Note

As of Red Hat Enterprise Linux Release 6.4, you can test the fence configuration for each node in a cluster with the fence_check utility. For information on this utility, see the fence_check(8) man page.

Fencing Configuration Examples

The following examples show a simple configuration with one fence method per node and one fence device per fence method:
The following examples show more complex configurations:

Note

The examples in this section are not exhaustive; that is, there may be other ways to configure fencing depending on your requirements.

Example 8.3. APC Fence Device Added to cluster.conf 


<cluster name="mycluster" config_version="3">
   <clusternodes>
     <clusternode name="node-01.example.com" nodeid="1">
         <fence>
         </fence>
     </clusternode>
     <clusternode name="node-02.example.com" nodeid="2">
         <fence>
         </fence>
     </clusternode>
     <clusternode name="node-03.example.com" nodeid="3">
         <fence>
         </fence>
     </clusternode>
   </clusternodes>
   <fencedevices>
         <fencedevice agent="fence_apc" ipaddr="apc_ip_example" login="login_example" name="apc" passwd="password_example"/>
   </fencedevices>
   <rm>
   </rm>
</cluster>
In this example, a fence device (fencedevice) has been added to the fencedevices element, specifying the fence agent (agent) as fence_apc, the IP address (ipaddr) as apc_ip_example, the login (login) as login_example, the name of the fence device (name) as apc, and the password (passwd) as password_example.

Example 8.4. Fence Methods Added to cluster.conf 


<cluster name="mycluster" config_version="3">
   <clusternodes>
     <clusternode name="node-01.example.com" nodeid="1">
         <fence>
            <method name="APC">
              <device name="apc" port="1"/>
             </method>
         </fence>
     </clusternode>
     <clusternode name="node-02.example.com" nodeid="2">
         <fence>
            <method name="APC">
              <device name="apc" port="2"/>
            </method>
         </fence>
     </clusternode>
     <clusternode name="node-03.example.com" nodeid="3">
         <fence>
            <method name="APC">
              <device name="apc" port="3"/>
            </method>
         </fence>
     </clusternode>
   </clusternodes>
   <fencedevices>
         <fencedevice agent="fence_apc" ipaddr="apc_ip_example" login="login_example" name="apc" passwd="password_example"/>
   </fencedevices>
   <rm>
   </rm>
</cluster>
In this example, a fence method (method) has been added to each node. The name of the fence method (name) for each node is APC. The device (device) for the fence method in each node specifies the name (name) as apc and a unique APC switch power port number (port) for each node. For example, the port number for node-01.example.com is 1 (port="1"). The device name for each node (device name="apc") points to the fence device by the name (name) of apc in this line of the fencedevices element: fencedevice agent="fence_apc" ipaddr="apc_ip_example" login="login_example" name="apc" passwd="password_example".

Example 8.5. cluster.conf: Multiple Fence Methods per Node


<cluster name="mycluster" config_version="3">
   <clusternodes>
     <clusternode name="node-01.example.com" nodeid="1">
         <fence>
            <method name="APC">
              <device name="apc" port="1"/>
             </method>
            <method name="SAN">
	      <device name="sanswitch1" port="11"/>
             </method>
         </fence>
         <unfence>
             <device name="sanswitch1" port="11" action="on"/> 
         </unfence>
     </clusternode>
     <clusternode name="node-02.example.com" nodeid="2">
         <fence>
            <method name="APC">
              <device name="apc" port="2"/>
            </method>
            <method name="SAN">
	      <device name="sanswitch1" port="12"/>
             </method>
         </fence>
         <unfence>
             <device name="sanswitch1" port="12" action="on"/> 
         </unfence>
     </clusternode>
     <clusternode name="node-03.example.com" nodeid="3">
         <fence>
            <method name="APC">
              <device name="apc" port="3"/>
            </method>
            <method name="SAN">
	      <device name="sanswitch1" port="13"/>
             </method>
         </fence>
         <unfence>
             <device name="sanswitch1" port="13" action="on"/> 
         </unfence>
     </clusternode>
   </clusternodes>
   <fencedevices>
        <fencedevice agent="fence_apc" ipaddr="apc_ip_example" login="login_example" name="apc" passwd="password_example"/>
        <fencedevice agent="fence_sanbox2" ipaddr="san_ip_example"
login="login_example" name="sanswitch1" passwd="password_example"/>
   </fencedevices>
   <rm>
   </rm>
</cluster>

Example 8.6. cluster.conf: Fencing, Multipath Multiple Ports


<cluster name="mycluster" config_version="3">
   <clusternodes>
     <clusternode name="node-01.example.com" nodeid="1">
         <fence>
            <method name="SAN-multi">
	      <device name="sanswitch1" port="11"/>
	      <device name="sanswitch2" port="11"/>
	    </method>
         </fence>
         <unfence>
             <device name="sanswitch1" port="11" action="on"/>
             <device name="sanswitch2" port="11" action="on"/>
         </unfence>
     </clusternode>
     <clusternode name="node-02.example.com" nodeid="2">
         <fence>
            <method name="SAN-multi">
	      <device name="sanswitch1" port="12"/>
	      <device name="sanswitch2" port="12"/>
            </method>
         </fence>
         <unfence>
             <device name="sanswitch1" port="12" action="on"/>
             <device name="sanswitch2" port="12" action="on"/>
         </unfence>
     </clusternode>
     <clusternode name="node-03.example.com" nodeid="3">
         <fence>
            <method name="SAN-multi">
	      <device name="sanswitch1" port="13"/>
	      <device name="sanswitch2" port="13"/>
            </method>
         </fence>
         <unfence>
             <device name="sanswitch1" port="13" action="on"/>
             <device name="sanswitch2" port="13" action="on"/>
         </unfence>
     </clusternode>
   </clusternodes>
   <fencedevices>
        <fencedevice agent="fence_sanbox2" ipaddr="san_ip_example"
login="login_example" name="sanswitch1" passwd="password_example"/> 
        <fencedevice agent="fence_sanbox2" ipaddr="san_ip_example"
login="login_example" name="sanswitch2" passwd="password_example"/> 
   </fencedevices>
   <rm>
   </rm>
</cluster>

Example 8.7. cluster.conf: Fencing Nodes with Dual Power Supplies


<cluster name="mycluster" config_version="3">
   <clusternodes>
     <clusternode name="node-01.example.com" nodeid="1">
         <fence>
            <method name="APC-dual">
              <device name="apc1" port="1"action="off"/>
              <device name="apc2" port="1"action="off"/>
              <device name="apc1" port="1"action="on"/>
              <device name="apc2" port="1"action="on"/>
             </method>
         </fence>
     </clusternode>
     <clusternode name="node-02.example.com" nodeid="2">
         <fence>
            <method name="APC-dual">
              <device name="apc1" port="2"action="off"/>
              <device name="apc2" port="2"action="off"/>
              <device name="apc1" port="2"action="on"/>
              <device name="apc2" port="2"action="on"/>
            </method>
         </fence>
     </clusternode>
     <clusternode name="node-03.example.com" nodeid="3">
         <fence>
            <method name="APC-dual">
              <device name="apc1" port="3"action="off"/>
              <device name="apc2" port="3"action="off"/>
              <device name="apc1" port="3"action="on"/>
              <device name="apc2" port="3"action="on"/>
            </method>
         </fence>
     </clusternode>
   </clusternodes>
   <fencedevices>
       <fencedevice agent="fence_apc" ipaddr="apc_ip_example" login="login_example" name="apc1" passwd="password_example"/>
       <fencedevice agent="fence_apc" ipaddr="apc_ip_example" login="login_example" name="apc2" passwd="password_example"/>
   </fencedevices>
   <rm>
   </rm>
</cluster>
When using power switches to fence nodes with dual power supplies, the agents must be told to turn off both power ports before restoring power to either port. The default off-on behavior of the agent could result in the power never being fully disabled to the node.