Hitachi

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


7.4.1 Starting a server in the wait state and then restarting jobs

The HA Monitor on the local host communicates with the HA Monitor on the remote host. An HA Monitor monitors the server status in both the active system and the standby system and determines the server status on the local host according to the server status on the remote host. Therefore, a server being started might be placed in a wait state before startup is completed. This subsection explains how to start and terminate a server that is in the wait state.

There are several types of server wait state and the start and termination methods differ depending on the type. When a server is placed in the wait state, a message to that effect is output at a specific interval. The following table describes the types of server wait states and the operating methods.

Table 7‒3: Types of server wait states and operating methods

Type

Text displayed by the monshow command

Description

Message

Start and termination methods

Restart wait state

*ONL*

The active server can be restarted after a failure has occurred on it. The status in effect until the restart is completed after a failure has occurred is called the restart wait state.

KAMN258-D

Start method: Start command provided by the program

Termination method: mondeact

Hot-standby wait state

ONL??

If a host reset is performed and hot standby processing occurs, HA Monitor switches the standby server in the standby system to the active server. When HA Monitor performs this switchover, it checks whether the active server in the active system (host where the failure occurred) has been terminated.

If HA Monitor determines that the active server has terminated, it starts the standby server in the standby system as the active server.

If HA Monitor cannot verify that the active server has terminated, the standby server in the standby system is placed in the wait state before it is actually started as the active server. The status in which the server in the standby system is waiting to be started as the active system is called the hot-standby wait state.

The server is also placed in the hot-standby switchover wait state if SCSI reservation for shared disk fails after host reset by hybrid fencing fails.

KAMN368-D

Start method: monact

Termination method: mondeact

Active server start wait state

*SBY*

If the status of the active server cannot be checked for a reason such as a monitoring path failure, the standby server waits until the active server has started. The status in which the standby server is waiting for the active server to start is called the active server start wait state.

KAMN238-D

Start method: monact

Termination method: mondeact

Resource server start wait state

If the resource server has not completed its startup processing, its child servers wait until the resource server has started. The status in which servers are waiting for startup of the resource server is called the resource server start wait state.

None

Start method: Executing monact after the resource server has started

Termination method: Executing mondeact after the resource server has terminated or started

Grouped-system switchover wait state

SBY??

If multiple servers are grouped together, you can specify the settings so that a grouped-system switchover is not performed, even if a failure occurs on one of the servers. If a failure occurs on a server that is set to not perform grouped-system switchover, the corresponding standby server is placed in the grouped-system switchover wait state.

If an active or standby server starts on any other host, the corresponding standby server returns to the status in which hot standby processing can be performed.

None

Start method: monact

Termination method: monsbystp

Start retry state

>ONL

If resource connection and server startup fail on the target host during hot standby processing but there is no available switchover host and the start retry function is used during hot standby processing, HA Monitor issues a restart instruction to the server. The status in effect from when the hot standby error occurs until server startup processing starts is called the start retry state.

KAMN296-I

Start method: Start command provided by the program

Termination method: mondeact

In the table, the Start and termination methods column provides the names of the applicable HA Monitor commands. The Text displayed by monshow command column indicates the server status that is displayed when the server and host status display command (monshow command) is executed. For details about the HA Monitor commands, see 9. Commands. For details about the server status transitions, see 4.1.3 Server status transitions.