Hitachi

For Linux(R) (x86) Systems HA Monitor Cluster Software


7.2.1 Starting

This subsection explains how to start the system, including the procedure from powering up the computer to reaching the system status in which servers are ready for the hot standby operation.

To start the system, you must start both the active system and the standby system. The following figure shows the procedure for starting the system.

Figure 7‒2: System start procedure

[Figure]

When the power has been turned on at both hosts and the servers have started, the two HA Monitors communicate with each other to check that they are ready for the hot standby operation. The KAMN252-I message is issued at this point.

In the case where host reset is performed with model HA8000xN or later:

If a management server is used as a failure management processor:

Before starting HA Monitor, you must start the management server and HA Monitor Connector must be started on the management server.

If HA Monitor is started first, the reset path configuration fails. Therefore, hot-standby switchover is not performed automatically if a host failure occurs. In such a case, use the following procedure to configure the reset path:

  1. Start the management server, HCSM, and HA Monitor Connector.

  2. Execute the reset path status display command (monrp command) on all hosts.

If a management server is not used as a failure management processor:

Before starting HA Monitor, you must start the services of HA Monitor Connector that manages the device on which the HA Monitor will operate.

If HA Monitor is started first, the reset path configuration fails. Therefore, hot-standby switchover is not performed automatically if a host failure occurs. In such a case, use the following procedure to configure the reset path:

  1. Start HA Monitor Connector.

  2. Execute the reset path status display command (monrp command) on all hosts.

    Important

    Do not terminate the management server and HA Monitor Connector while HA Monitor is operating. If you terminate it, the HA Monitors on all monitoring-target hosts detect a reset path failure. If system operation continues in that status, it is impossible to reset a host on which a host failure occurs. Therefore, hot-standby switchover cannot be performed automatically. In such a case, restart the management server or HA Monitor Connector, and then execute the reset path status display command (monrp command) on all hosts.

The following subsections explain how to start HA Monitor, the resource server, and servers, as shown in Figure 7-2.

Organization of this subsection

(1) Starting HA Monitor

To start HA Monitor manually, you execute the HA Monitor start command (monstart). After you have started HA Monitor, make sure that the KAMN002-I message indicating that the HA Monitor startup processing is completed is displayed.

When HA Monitor is to be started automatically, it is started by the OS after the kernel has started.

When HA Monitor is first installed, it is set to be started automatically. To change the HA Monitor start method, execute the HA Monitor environment setup command (monsetup command). For details about the HA Monitor environment setup command (monsetup command), see 9.22 monsetup (specifies HA Monitor environment settings) in 9. Commands.

(2) Starting a resource server

A resource server is started automatically when a server that has a resource server as its parent server is started. You specify a parent server in the server environment definition.

A resource server is terminated automatically by HA Monitor when all servers using the resource server have terminated.

If you are starting a resource server manually for a reason such as a shared resource connection error, make sure that the KAMN002-I message indicating that the HA Monitor startup processing is completed is displayed and then execute the resource server start command (monresbgn command).

(a) Action to be taken if the system is placed in wait for startup of the active server

When the resource server is started, the system might be placed in wait for startup of the active server. For cases in which the system is placed in wait for startup of the active server, see 4.1.1 HA Monitor's control of server startup and 4.5.3 Controlling the server start and termination (multi-standby).

For details about the action to be taken if the system is placed in wait for startup of the active server, see 7.4.1 Starting a server in the wait state and then restarting jobs.

(3) Starting servers

The servers in both the active system and the standby system must be started. Before you start servers manually, make sure that the KAMN002-I message indicating that the HA Monitor startup processing is completed is displayed.

To start a server in the server mode, you execute the start command provided by the program. When you start a server, it creates an environment for the hot standby operation by communicating with HA Monitor during the server initialization process. A server started in the standby system becomes the standby server and prepares for any failure that might occur in the active system.

To start a server in the monitor mode, you use the monitor-mode server start command (monbegin command). A server started in the standby system becomes the standby server and prepares for any failure that might occur in the active system.

After you have started a server, make sure that the KAMN251-I message indicating that the server startup processing is completed is displayed. The KAMN251-I message is displayed for each server. If multiple servers are run on the same host, this message is displayed for each of them.

(a) Points to be noted when servers are grouped together

When multiple servers are grouped together, all the servers in the group are started. The servers are started in the order specified by the operator. After you have started grouped servers, make sure that the KAMN251-I message is displayed for all of the servers in the group.

(b) Action to be taken if the system is placed in wait for startup of the active server

When servers are started, the system might be placed in wait for startup of the active server. For cases in which the system is placed in wait for startup of the active server, see 4.1.1 HA Monitor's control of server startup and 4.5.3 Controlling the server start and termination (multi-standby).

For details about the action to be taken if the system is placed in wait for startup of the active server, see 7.4.1 Starting a server in the wait state and then restarting jobs.