3.2.7 Performing start retries during hot standby processing
If resource connection and server startup fail on the target host during hot standby processing and there is no available switchover host, a hot standby error occurs and the job stops.
HA Monitor enables you to retry restarting the server on the target host, thereby avoiding a hot standby error due to a transient cause.
The following describes the servers that can use the start retry function and the start retry range:
-
Server-mode servers other than TP1/EE
In the event of a hot standby error, restart is retried either until hot standby processing is completed or for as many times as specified.
-
Servers in the monitor mode and resource servers
During hot standby processing, restart is retried either until hot standby is enabled or for as many times as specified.
If the server on the target host fails to start, server restart is attempted in the following cases:
-
Server in the server mode failed to start.
-
Server in the monitor mode failed to start or a server monitoring command detected a failure.
-
The function for canceling server startup in the event of a shared resource connection error is being used and the shared resource connection processing failed.
-
The function for canceling server startup in the event of a shared resource inheritance timeout error is being used and the shared resource connection processing timed out.
-
SCSI reservation for shared disk is being used as a shared disk data protection method and the reservation failed.
If another host's active server starts while start retries are underway on the local host, the server on the local host will be started as a standby server.
In a grouped-system switchover configuration, start retries are performed for each server. If a server group contains a server engaged in start retry processing and another server in the same server group successfully switches over to the remote host, the start reties are canceled.
- Organization of this subsection
(1) Overview of start retries during hot standby processing
If the server on the target host fails to start, HA Monitor uses the following procedure to retry server startup:
-
HA Monitor places the server in the start retry state (>ONL) and waits for the server to be restarted.
-
The server is restarted (the server must be set to be restarted).
-
When HA Monitor accepts the server restart, it starts the server as the active server.
When start retries are performed, the flow of processing, including messages, user commands, and JP1 events, is the same as during normal active server startup, not during hot standby processing.
The following figure shows the flow of start retry processing.
The following figure shows the flow of start retry processing when the multi-standby function is used.
(2) Required environment settings
You must specify the switch_error operand in the server environment definition to perform start retries during hot standby processing.
In server mode, TP1/Server Base and HiRDB must also be set up. For details, see the switch_error operand in 8.4.1 Server environment definition (servers).
For a server in the monitor mode or a resource server, you can specify a retry count in the switch_retry operand in the server environment definition.