[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

[Linux-cluster] Active-Active configuration of arbitrary services



We are running RHCS on RHEL 4.5 and have a basic 2-node HA cluster configuration for a critical application in place and functional. The config looks like this:

 

<?xml version="1.0"?>
<cluster config_version="16" name="routing_cluster">
        <fence_daemon post_fail_delay="0" post_join_delay="10"/>
        <clusternodes>
                <clusternode name="host1" votes="1">
                        <fence>
                                <method name="1">
                                        <device name="manual" nodename="host1"/>
                                </method>
                        </fence>
                </clusternode>
                <clusternode name="host2" votes="1">
                        <fence>
                                <method name="1">
                                        <device name="manual" nodename="host2"/>
                                </method>
                        </fence>
                </clusternode>
        </clusternodes>
        <cman dead_node_timeout="10" expected_votes="1" two_node="1"/>
        <fencedevices>
                <fencedevice agent="fence_manual" name="manual"/>
        </fencedevices>
        <rm>
                <failoverdomains>
                        <failoverdomain name="routing_servers" ordered="1" restricted="1">
                                <failoverdomainnode name="host1" priority="1"/>
                                <failoverdomainnode name="host2" priority="2"/>
                        </failoverdomain>
                </failoverdomains>
                <resources>
                        <script file="/etc/init.d/rsd" name="rsd"/>
                        <ip address="123.456.78.9" monitor_link="1"/>
                </resources>
                <service autostart="1" domain="routing_servers" name="routing_daemon" recovery="relocate">
                        <ip ref="123.456.78.9"/>
                        <script ref="rsd"/>
                </service>
        </rm>
</cluster>

 

The cluster takes about 15-20 seconds to notice that the daemon is down and migrate it to the other node. However, due to slow migration and startup time, we now require the daemon on the secondary to be active and only transfer the VIP in case it aborts on the primary.

 

I have found the cluster suite documentation to be lacking, particularly in its reliance on GUI configuration tools without reference or explanation to the underlying command-line tools. I managed to figure out most of it by myself (we don’t run GUIs on servers), but I’m not sure how to keep both services active, using the init script for status, while only moving the VIP. Is this possible? Am I overlooking the obvious?

 

Thanks in advance for suggestions/help!

 

Glenn


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]