Cluster Maintenance

Starting a Cluster

Run the following command to start a cluster:

smit clstart

Then, configure the following parameters:

• Start Cluster Services on these nodes

  Indicates the two cluster nodes to be started upon the cluster startup. The two nodes can be started one by one or at the same time.

• Startup Cluster Information Daemon

  Indicates whether to start clinfoES upon the cluster startup. If this parameter is set to false, you cannot run the #/usr/sbin/cluster/clstat -a command to view cluster running status.

Check whether any errors occur during the cluster startup.

If any errors occur, stop the cluster first and then rectify faults based on error messages. Then, start the cluster again.

Stopping a Cluster

Run the following command to stop a cluster:

smit clstop

Then, configure the following parameters:

• Stop Cluster Services on these nodes

  Indicates the cluster nodes to be stopped.

• Select an Action on Resource Groups

  Indicates the mode to stop a cluster.

  Possible values are bring resource groups offline (graceface), move resource groups (takeover), and unmanage resource groups (forced). The meaning of each value is as follows:

  • bring resource groups offline

    Stops the cluster on this node. The peer node is not affected.

  • move resource groups

    Stops the cluster on this node. The peer node takes resources from this node.

  • unmanage resource groups

    Forcibly stops HA on this node without releasing any resources. The peer node is not affected.

Checking Cluster Status

The cluster status includes the cluster process status and cluster service status. Perform the following steps to check the cluster status:

1. Check the cluster process status on nodes. The command syntax is as follows:

   lssrc -g cluster

   The command output is as follows:

   Figure 10-1 Cluster process status
   

2. Check the cluster service status on nodes. The command syntax is as follows:

   #/usr/sbin/cluster/clstat -r 2 -a

   In this command, 2 indicates the display of current status every 2 seconds.

   The command output is as follows:

   Figure 10-2 Cluster service status
   

   In the output, the service IP address and the resource group of the cluster are on node ibm31 and the node is online, which indicate that the cluster is in normal state.

Cluster Switchover

Perform as follows to switch services between two nodes:

Run the smit hacmp command on the host and then choose:

System Management (C-SPOC) > HACMP Resource Group and Application Management > Move a Resource Group to Another Node