Attentions

There are some points user should pay more attention.

  1. On every HA node, there should be scripts for start, stop and restart HA software in directory /etc/HAscript/. They are:

               startha:     start  heartbeat and mon

                        startha.jpg (50454 bytes)

                        /usr/lib/heartbeat/heartbeat start : Start the heartbeat. It may change to /var/rc.d/init.d/heartbeat start.

                      mon &  : run mon in back ground style.  mon is under directory of /usr/bin

                       User can add other command in this script , such as httpd. This command would be run at the same time.

              stopha:    stop heartbeat and mon

                         stopha.jpg (50454 bytes)

              restartha:    stop heartbeat and mon , then start them   

                        restartha.jpg (65446 bytes)

      2.   For HA application:

                alert script is named by Appname.run , under directory /usr/lib/mon/alert.d/.

                monitor script is named by Appname.monitor , under directory /usr/lib/mon/mon.d/.

                Application start script named by Appname, under directory /etc/rc.d/init.d/.

      3.  HA application monitor script:

                        monitor.jpg (394734 bytes)

              This is the test.monitor script under directory /usr/lib/mon/mon.d/. HA application monitor would be just like this. ClusterProbe GUI would help user to create monitor script for HA application. It would replace underline with HA application keywords.

              User should make sure that this monitor would be getting the right result. That means, if the HA application exists, the result of this script would be "yes" and echo "yes" to standard output. Or it would echo "no" and the result would be "no" when it does not exist. 

             User also should change the mode of this script to runnable.

      4.  On ClusterProbe Server, Heartbeat rpm files and mon rpm files should be under directory /etc/HAscript/tmp/rpm/. And on HA nodes, rpm files copied from ClusterProbe server would saved under directory /HArpm/.

      5.  In file ha.cf and haresources, node name should be full name, including network name.