Scalable Database Server, HiRDB Version 8 System Definition

[Contents][Index][Back][Next]

2.3.3 Operands related to HiRDB startup

9) pd_mode_conf = AUTO | MANUAL1 | MANUAL2
Specifies the HiRDB startup method.
AUTO:
HiRDB is automatically started. The automatic startup method automatically starts HiRDB when the OS is started.
However, you must manually start HiRDB in the following cases. In the manual startup method, you must execute the pdstart command to start HiRDB.
  • Normal start following normal termination (without restarting the OS)
  • Restart following planned termination (without restarting the OS)
  • Restart following forced termination
MANUAL1:
HiRDB is manually started. However, after abnormal termination, HiRDB is automatically started.
MANUAL2:
Manual startup.
Specification guidelines
  • To restart HiRDB (or the unit) automatically after HiRDB (or the unit) has terminated abnormally, specify AUTO or MANUAL1.
  • If you use the system switchover facility, specify MANUAL1 or MANUAL2. Which of these modes is more appropriate depends on how the system switchover facility is used. For the methods of operating the system switchover facility, see the HiRDB Version 8 System Operation Guide.
  • The startup mode is determined by the combination of the previous termination mode (normal termination, forced termination, planned termination, or abnormal termination) and the startup method specified by this operand, as shown in the following table:
    pd_mode_conf specification Previous termination mode Startup mode Startup method
    AUTO Normal termination Normal startup Automatic startup or Manual startup1
    Planned termination Restart2
    Forced termination Restart2 Manual startup
    Abnormal termination Restart Automatic startup
    MANUAL1 Normal termination Normal startup Manual startup
    Planned termination Restart2
    Forced termination Restart2
    Abnormal termination Restart Automatic startup
    MANUAL2 Normal termination Normal startup Manual startup
    Planned termination Restart2
    Forced termination Restart2
    Abnormal termination Restart2
    1 Automatic startup is used only when the OS is started. For normal start following normal termination and restart following planned termination (without restarting the OS), manual startup is used.
    2 The dbdestroy option of the pdstart command can be used for a forced startup. However, when forced startup is used, the HiRDB database is not recovered automatically and must be recovered by the HiRDB administrator.
  • If you use Real Time SAN Replication based on the log-only synchronous method, the startup mode is always set to normal startup regardless of the value of this operand or the previous termination mode. In the case of a log application site, the changeable system definition operands are only those that can be changed during restart.
Note
  • To use automatic startup (pd_mode_conf=AUTO) in a HiRDB/Parallel Server, try to start all units within 20 minutes after starting the first unit. If all units are not started within 20 minutes, the HiRDB startup process is cancelled. You can use the pd_reduced_check_time operand to change this 20-minute time limit.
    Note that this time limit does not apply when you are restarting a unit after it or the OS was abnormally terminated.
  • If you use Real Time SAN Replication, AUTO cannot be specified. If AUTO is specified, HiRDB startup processing results in an error.
  • If you use Real Time SAN Replication based on the log-only synchronous method, MANUAL2 is assumed even when MANUAL1 is specified in this operand at the log application site.

10) pd_system_complete_wait_time = pdstart-command-completion-wait-time
[Figure]<unsigned integer>((610-3600))<<610>>(Seconds)
Specifies the amount of time to wait for completion of the pdstart command's processing. This operand needs to be specified only when it is necessary to provide a longer wait time than the default value of 610 seconds (10 minutes and 10 seconds). If startup processing is not completed within the specified number of seconds after the pdstart command was entered, HiRDB outputs the KFPS05078-I message and returns the pdstart command with an error.
Even though the pdstart command is returned with an error, startup processing continues and HiRDB is started. Therefore, this operand should be specified only in the following cases:
  • When the KFPS05078-I message is the monitoring target
  • When another operation is to be performed automatically after the pdstart command terminates normally
Application criterion
Normally, this operand need not be specified. This operand is applied when the KFPS05078-I message is output and the pdstart command is returned with an error during HiRDB startup processing.

11) pd_start_time_out = HiRDB-start-preparation-maximum-wait-time
[Figure]<unsigned integer>((1-1440))<<15>>(minutes)
Specifies the maximum amount of time to wait for completion of HiRDB start preparation after pdstart command input. If HiRDB start preparation is not completed within the time specified in this operand after pdstart command input, the KFPS01861-E message (reason code = TIMEOUT) is output in the pdstart command input window.
Specification guidelines
Normally, you need not specify this operand. In an ordinary environment, the HiRDB start preparation time is at most 3 to 6 minutes, and therefore the default value [15 minutes] of this operand is sufficient for the maximum wait time. In rare cases, HiRDB start preparation may take more than 15 minutes because of factors such as server machine performance, disk size, and memory load. In this case, the KFPS01861-E message (reason code = TIMEOUT) is output in the pdstart command input window. If this occurs, specify a value greater than 15 minutes for this operand.
Reference note
Based on the time at which the following message was output, you can compute the wait time for the completion of the HiRDB startup preparation process and the wait time for the completion of the HiRDB startup:
[Figure]

12) pd_term_watch_count = maximum-number-of-successive-abnormal-terminations

[Figure]<unsigned integer>((1-3))<<3>>

Specifies the maximum number of times in a row that abnormal termination can occur. If HiRDB terminates abnormally during HiRDB (or unit) restart attempts for the number of times in a row specified in this operand, restart processing is canceled. For example, if 2 is specified here, restart processing is canceled after two HiRDB restarts in a row terminate abnormally.

The HiRDB restart attempts can be either automatic or manual, depending on the value specified in the pd_mode_conf operand.

When HiRDB terminates abnormally after the number of abnormal terminations specified in this operand has been reached, the KFPS00715-E message is output. In this case, remove the cause of the error and then execute the pdrpause command in the unit indicated in the KFPS00715-E message.