This section explains how to change the operation state of the PRIMECLUSTER system. The operations include starting and stopping a cluster application.
The procedure for starting a cluster application is described below.
Operation Procedure:
On the RMS tree in the RMS main window, right-click the cluster application to be started, and select Online from the pop-up menu that is displayed.
The cluster application will start.
Information
You can also display the pop-up menu by right-clicking the target icon in an RMS graph or the RMS cluster table. For details on RMS graphs and the RMS cluster table, see "7.3.5 Viewing Detailed Resource Information" and "7.3.3 Concurrent Viewing of Node and Cluster Application States."
Note
To start a cluster application manually, check that the cluster application and resources under it are stopped on all the nodes other than the node on which the cluster application is to be started. You can check whether they are stopped by the Offline or Standby state. With the state other than Offline or Standby, they may be running. In this case, stop them and then start the cluster application on the target node.
The procedure for stopping a cluster application is described below.
Operation Procedure:
On the RMS tree in the RMS main window, right-click the cluster application to be stopped, and select Offline from the displayed pop-up menu.
The cluster application will stop.
Information
You can also display the pop-up menu by right-clicking the target icon in an RMS graph or the RMS cluster table. For details on RMS graphs and the RMS cluster table, see "7.3.5 Viewing Detailed Resource Information" and "7.3.3 Concurrent Viewing of Node and Cluster Application States."
The procedure for switching a cluster application is described below.
Operation Procedure:
Right-click on the application object and select the Switch menu option.
A pull-down menu appears listing the available nodes for switchover.
Select the target node from the pull-down menu to switch the application to that node.
Information
You can also display the pop-up menu by right-clicking the target icon in an RMS graph or the RMS cluster table. For details on RMS graphs and the RMS cluster table, see "7.3.5 Viewing Detailed Resource Information" and "7.3.3 Concurrent Viewing of Node and Cluster Application States."
The procedure for bringing a Faulted cluster application to available state is described below.
Operation Procedure:
Right-click on the cluster application object in the RMS tree, and select Clear Fault.
Information
You can also display the pop-up menu by right-clicking the target icon in an RMS graph or the RMS cluster table. For details on RMS graphs and the RMS cluster table, see "7.3.5 Viewing Detailed Resource Information" and "7.3.3 Concurrent Viewing of Node and Cluster Application States."
A node becomes the Wait state during state transition. The procedure for clearing the Wait state of a node is described below.
Operation Procedure:
Check that the node in the Wait state has been stopped. If not, stop the node manually.
Check that the CF state is DOWN in the CF main window. If the CF state is LEFTCLUSTER, clear LEFTCLUSTER in the CF main window and make sure the node state is changed from LEFTCLUSTER to DOWN.
If the Wait state of the node has not been cleared after performing 2, right-click on the system node in the RMS graph and select the "Clear Wait & shutdown (hvutil -u)" from the menu.
Note
If you clear the Wait state of a system node manually, RMS and CF assume that you have already checked that the target node had stopped. Therefore, if you clear the Wait state when the node has not been stopped, this may lead to the data corruption.
Information
You can also display the pop-up menu by right-clicking the target icon in an RMS graph or the RMS cluster table. For details on RMS graphs and the RMS cluster table, see "7.3.5 Viewing Detailed Resource Information" and "7.3.3 Concurrent Viewing of Node and Cluster Application States."
The maintenance mode is a specific mode to temporarily restrict a specified cluster application switching.
If a cluster application becomes maintenance mode, it cannot be switched.
Note that cluster nodes and resources are monitored during maintenance mode. In this case, when the resource state is changed, the resource state of the cluster application that is viewed on the RMS tree is also changed.
If the state of a cluster application resource has changed while in maintenance mode, since switching is not carried out, it becomes a state in which consistency with the resource registered in the cluster application is collapsed. (Example: Some resources are in the Offline state while others are in the Online state.) Therefore, before exiting the maintenance mode, it is necessary to revert the resource state of the cluster application to the same state as before starting the maintenance mode.
For using maintenance mode, see "7.4 Using maintenance mode" in "PRIMECLUSTER Reliant Monitor Services (RMS) with Wizard Tools Configuration and Administration Guide."
Note
Please note the following for using maintenance mode.
Perform maintenance mode to the cluster application of the standby operation containing resources for which the maintenance is necessary.
Since the resources for which the maintenance is necessary during the operation are not contained, it is not necessary to make the cluster application of the scalable operation into maintenance mode.
To start maintenance mode, a cluster application must be in the Online, Standby, or Offline state on all the nodes.
To exit maintenance mode, a cluster application and each resource must be returned in the same state before starting maintenance mode.
Do not stop RMS or the system with cluster applications in maintenance mode. Be sure to exit maintenance mode of all cluster applications before stopping RMS or the system.
Use maintenance mode only when applicable products are specified in the environment that uses PRIMECLUSTER products.
When the cluster application that includes Cmdline resource that sets the NULLDETECTOR flag is in maintenance mode, the script that was set to the Cmdline resource must correspond to the maintenance mode. For details, see "6.11.2.1.4 Notes When Setting the NULLDETECTOR Flag."
For details, see "7.4.2 Maintenance mode operating notes" or "2.1.7.1 Restrictions during maintenance mode" in "PRIMECLUSTER Reliant Monitor Services (RMS) with Wizard Tools Configuration and Administration Guide."
