The pdstart command starts a HiRDB system, unit, or server. This command is also used to initiate a restart.

HiRDB administrator

(a) HiRDB single server configuration

 
 pdstart [{-i|-r[ -t]|-l|dbdestroy}]
 

(b) HiRDB parallel server configuration

Starting the HiRDB system

pdstart [{-i|-r[ -t]|-l|dbdestroy}]

Starting a unit

pdstart {-x host-name|-u unit-identifier}[{-r[ -t]|-l|dbdestroy}]

Starting a server

pdstart -s server-name

Starting the system by resetting the front-end server from its SUSPENDed status

pdstart -a [-s front-end-server-name]

Starting a unit from a unit

pdstart -q [{-r[ -t]|-l}]

Starting the alternate system when using the standby-less system switchover (1:1) facility (placing the alternate system in standby status)

pdstart -q -c

Starting a server on the specified target unit when using the standby-less system switchover (effects distributed) facility

pdstart [{-x host-name|-u unit-identifier}] -s server-name

Forcibly starting the unit for the system manager that terminated abnormally when the HiRDB parallel server configuration was started by the pdstart -r command

pdstart -R [-t]

(a) {-i|-r[ -t]|-l|dbdestroy}

Specifies the startup mode for the HiRDB system. When this option is omitted, the HiRDB system performs normal startup if the previous termination was normal termination or performs a restart if the previous termination was forced or abnormal termination. In the case of a restart, database recovery processing is performed.

-i

If databases have been initialized by the database initialization utility (pdinit), specifies that the databases are to be reinitialized.

-r

Specifies that the master directory RDAREA is to be backed up by the database copy utility (pdcopy) (-M x specified). Also specifies that if HiRDB will not start due to an error in the master directory RDAREA, the database recovery utility (pdrstr) is to be executed.

-t:

Specifies that the shared resources subject to system switchover are to be activated in advance, before the database copy utility (pdcopy) or database recovery utility (pdrstr) is executed. This option is applicable when an HA monitor is used as cluster software and the server-mode system switchover facility is employed.

When you specify this option, you must also specify the -r or -R option.

-l:

When Real Time SAN Replication based on the log-only synchronous method is used, specifies that HiRDB is to be started at the log application site.

At the log application site, the system log information sent from the transaction execution site is input to keep the database up-to-date at the log application site. Therefore, the database cannot be accessed from the log application site.

Rules

1. When Real Time SAN Replication based on the log-only synchronous method is used, the command with the -l option specified can be executed from the log application site. If it is executed from the transaction execution site, an error results.
2. When log application processing is executed with the -l option specified, the system switchover facility cannot be used. Even if HiRDB terminates abnormally due to a failure during log application processing, system switchover does not take place.
   Also note that the pdstart command with the -l option specified cannot be executed at the secondary system. If it is executed, the KFPS05813-E message is displayed.
   
3. When log application processing is executed with the -l option specified, neither reduced activation nor a recovery-unnecessary front-end server can be used. This is because the HiRDB at the log application site cannot be started when there are units that have not started.
4. In the case of a HiRDB parallel server configuration with multiple units to which Real Time SAN Replication based on the log-only synchronous method is applied, the files for storing the site status are placed only in the unit containing the system manager. In other units, the site status is determined by instructions from the system manager unit.
   Therefore, in the case of a HiRDB parallel server configuration to which Real Time SAN Replication based on the log-only synchronous method is applied, when you independently start a non-system manager unit as a transaction execution site, do not specify the -l option.
   On the other hand, to start the unit independently as a log application site, make sure that you specify the -l option.
   The following shows combinations of site status and pdstart command options when a non-system manager unit is started independently:
   

   Site status	Unit to be started	pdstart command's option			Combination of options
   		-u or -x	-q	-l
   Transaction execution	System manager	Y	N	N	Combinable
   		Y	N	Y	Not combinable
   	Other than system manager	N	Y	N	Combinable
   		N	Y	Y	Not combinable
   Log application	System manager	Y	N	N	Not combinable
   		Y	N	Y	Combinable
   	Other than system manager	N	Y	N	Not combinable
   		N	Y	Y	Combinable

   Legend:
   Y: Specified
   N: Omitted
   

dbdestroy

Specifies that HiRDB is to be restarted by forced restart because it would not restart. When HiRDB is restarted with this option specified, all RDAREAs (including system RDAREAs) are destroyed.

(b) {-x host-name|-u unit-identifier} [-r[ -t]|-l|dbdestroy]

Specifies that a unit is to be started.

-x host-name ~<identifier> ((1-32))

When a unit is to be started, specifies the name of the host containing the unit. This option cannot be specified together with the -i option.

When the standby-less system switchover (effects distributed) facility is used, the command starts the host and guest BESs on the specified host's unit. The command starts the server specified as the running system by the cluster software as the running system and places the server specified as the standby system in acceptable status. The server started as the running system is in acceptable status when viewed from another unit in the HA group.

-u unit-identifier ~<identifier> ((4 characters))

Specifies the identifier of the unit to be started. Note that the unit identifier of the unit that contains the system manager cannot be specified. Additionally, this option cannot be specified together with the -i option.

When the standby-less system switchover (effects distributed) facility is used, the command starts the host and guest BESs on the specified unit.

The command starts the server specified as the running system by the cluster software as the running system and places the server specified as the standby system in acceptable status. The server started as the running system is in acceptable status when viewed from another unit in the HA group.

For details about -r[ -t], -l, and dbdestroy, see the description of {-i|-r[ -t]|-l|dbdestroy}.

(c) -s server-name ~<identifier> ((1-8))

When a server is to be started, specifies the name of the server.

Notes when the standby-less system switchover (effects distributed) facility is used:

When the -u option is omitted, the command starts the specified servers on all active units in the HA group. Note that only the one server that was specified as the running system by the cluster software is started as the running system and that all other servers are started as being in acceptable status.

When the -u option is specified, the command starts the specified servers on that unit. In this case, the command starts the server specified as the running system by the cluster software as the running system and the server specified as the standby system as being in acceptable status. When viewed from other units in the HA group, the server started as the running system is in acceptable status.

For details about the standby-less system switchover (effects distributed) facility, see the HiRDB Version 9 System Operation Guide.

(d) -a [-s front-end-server-name]

Restarts a front-end server after resolving the cause of its SUSPEND status (waiting for the recovery of the data dictionary RDAREA or the starting of the dictionary server). Regardless of whether the -s option is specified, this option starts all the front-end servers in which an error occurred.

(e) -q [{-r[ -t]|-l}]

Specifies that the unit is to be started in a HiRDB parallel server configuration. When the standby system switchover facility is used, specify this option to start the standby unit.

When the -q option is specified, the host corresponding to the unit to be started (host with the -x or -c operand specified in the pdunit statement in the system common definition) is directly logged into and the command is executed.

If the -q option is specified, you must start all units before the pd_reduced_check_time operand value in the system definition is reached (when omitted, 20 minutes). If the command cannot start any of the units, HiRDB startup processing is cancelled.

For details about -r[ -t] and -l, see the description of {-i|-r[ -t]|-l|dbdestroy}.

(f) -q -c

When the standby-less system switchover (1:1) facility is used, specifies that the alternate system is to be placed in standby status. Specify this option in the following cases:

• If planned switchover is executed using a cluster software command, the alternate system is stopped. In such a case, use the pdstart -q -c command to place the alternate system in standby status as a safeguard in the event of an error.
• The alternate system is stopped if you stop the normal BES unit with a command such as pdstop -u and restart the normal BES unit with the pdstart -q command after the alternate system of the corresponding alternate BES unit has stopped. In such a case, use the pdstart -q -c command to place the alternate system in the standby status as a safeguard in the event of an error.

To execute the pdstart -q -c command, directly log onto the host of the alternate system that is to be placed in standby status (host specified in -x of the pdunit operand in the system definition).

(g) -R [-t]

This option is used when a HiRDB parallel server configuration terminated abnormally due to a failure on the unit for the system manager after the HiRDB parallel server configuration had been started by the pdstart -r command. For details about using this option to recover a HiRDB parallel server configuration, see Actions to take if an error occurs during a startup using the pdstart -r command in the HiRDB Version 9 System Operation Guide.

For details about -t, see the description of {-i|-r[ -t]|-l|dbdestroy}.

When the pdstart -R command is executed, HiRDB forcibly starts the unit for the system manager and then forcibly terminates all units that were running at the time the command executed. When the pdstart -R command is executed, the HiRDB processing depends on the system configuration and the status of each unit. The table below describes the HiRDB processing and return code that depends on the system configuration and unit status.

Legend:

--: No relation

1. The pdstart command can be executed to start a HiRDB system only while the HiRDB is shut down. To start a unit or server, this command can be executed only while the system manager is active.
2. The pdstart command must be executed on the server machine that contains the single server or the utility special unit or the server machine where the system manager is located.
3. If the system manager terminates abnormally in a HiRDB parallel server configuration, the server must be restarted with the pdstart command (with no options specified).
4. When the -i option is specified, any operation command or any utility command other than for the database initialization utility can be executed after the database initialization utility has terminated normally and the HiRDB system has been placed in startup completed status. The HiRDB system does not check a utility or operation command executed before the database initialization utility has terminated. If executed, subsequent HiRDB system operations cannot be guaranteed.
5. When you specify the -i option, make sure that all units are inactive. If any units are running when the pdstart -i command executes, startup of those units might not be recognized. Use the pdls -d ust command to check the status of each unit. If any unit is running, stop it with the pdstop -z command.
6. To place the alternate system in standby status when the standby-less system switchover (1:1) facility is used, you cannot specify the -i option in the pdstart command.
7. When the -r option is specified, the HiRDB system or unit is started only in order to execute the database copy utility or the database recovery utility. Therefore, when the HiRDB system or unit is started by the pdstart command with the -r option specified, no utility other than the database copy utility or the database recovery utility and no command other than pdls or pdstop should be executed. If executed, subsequent HiRDB system operations cannot be guaranteed.
8. When the dbdestroy option is specified, all RDAREAs (including system RDAREAs) updated since entry of the pdclose command or since the previous HiRDB startup are destroyed. When a HiRDB forced startup is performed with this option specified, it is necessary to either use the database recovery utility to recover all RDAREAs destroyed or to reload the entire database.
   For details about forced startup by specifying the dbdestroy option and for recovery of RDAREAs, see the HiRDB Version 9 System Operation Guide.
   
9. When you specify the dbdestroy option, make sure that all units are inactive. If any units are running when the pdstart dbdestroy command executes, startup of those units might not be recognized. Use the pdls -d ust command to check the status of each unit. If any unit is running, stop it with the pdstop -z command.
10. If you specify the -r option when the system switchover facility is being used, before you execute the pdstart command, you must activate the disk and start the IP address in a unit to which the standby-less system switchover (effects distributed) facility is not applied. In this case, the action depends on the -q option specification in the pdstart command as described below:
    
    • When the -q option is omitted
      Check the disk connection status and IP address operating status, and then activate the disk and start the IP address in the primary system.
      
    • When the -q option is specified
      Check the disk connection status and IP address operating status, and then activate the disk and start the IP address in the system where the pdstart command is executed.
      
11. When you use the pdstart -u command in a HiRDB parallel server configuration to start a unit that uses the system switchover facility without IP address inheritance, the primary unit is started as the running system.
    If you want to start the secondary unit as the running system, execute the pdstart -q command on the secondary unit.
    In the case of a unit to which the system switchover facility in the server mode is applied, and which uses the HA monitor as cluster software, if you start the secondary unit with the pdstart -u command, the unit is placed in wait status until the server in the running system starts.^# To start the secondary unit as the running system, execute the HA monitor's monact command.
    

    #: This status is displayed as *SBY* when the HA monitor's monshow command is executed.

12. If you use the system switchover facility and execute the pdstart -r command, activate the disk in advance and start the IP address in the target unit for system switchover (except for a unit to which the standby-less system switchover (effects distributed) facility is applied).
    The option described here is not required when the HA monitor is used as cluster software, the system switchover facility is applied in the server mode, and the pdstart -r -t object property is specified.
    For a HiRDB parallel server configuration, the HiRDB startup method depends on whether or not all units are to be started in the primary system, as described below. If the primary system cannot be started for a reason such as a server machine failure, start each unit separately as described under When some or all units are started in the secondary system below.
    
    • When all units are started in the primary system
      Check the disk connection status. If the disk in the primary system is not active, activate it. In the case of a unit that inherits IP addresses from the system switchover facility, make sure that the IP address is active in the primary system. If the IP address is not active, start it in the primary system. Check all primary systems to make sure that the disk is active and that the IP address has started, and then execute the pdstart -r command on the unit containing the system manager.
      
    • When some or all units are started in the secondary system
      Check the disk connection status. If the disk is not active in the system to be started, activate the disk in that system. In the case of a unit that inherits IP addresses from the system switchover facility, make sure that the IP address is active in the system that is to be started. If the IP address is not active, start it in that system. Make sure that the disk is active in all units and that the IP address has started, and then execute the pdstart -q -r command on each unit.
      

1. The following are the pdstart command's return codes:
   0: Normal termination
   4: A message was displayed because of a timeout.
   8: Abnormal termination (an invalid option was specified)
   
2. The result of the pdstart command can be checked by the pdls command (with -d prc or -d svr specified).
3. If the pdstart command terminates abnormally or the pdstart command's process is terminated by a method such as entering the OS's kill command, you must use the pdls -d svr command to check the status of HiRDB. If the HiRDB unit and all user servers are running (ACTIVE), HiRDB has started. If it takes more time than usual for the HiRDB unit or user servers to be placed in ACTIVE status, forcibly terminate HiRDB (with the pdstop command in which the -f or -z option is specified), and then re-execute the pdstart command. If the KFPS01853-W message displays unit state not ONLINE after execution of the pdls -d svr command, use the pdls -d ust command to check the unit status. If the HiRDB unit remains in STARTING status for a long time, forcibly terminate HiRDB (with the pdstop command in which the -f or -z option is specified), and then re-execute the pdstart command.
4. If you are using Real Time SAN Replication based on the log-only synchronous method, and if the pdstart command has been executed with the -i, -r, or dbdestroy option specified at the transaction execution site, execute the preparations for log application.
5. The following are possible causes for the KFPS01801-E message (reason code=SETUP) and the actions to be taken; this message may be displayed during execution of the pdstart command:
   

   Causes:

   The HiRDB directory set in the PDDIR environment variable has not been registered into the OS by the pdsetup command.

   The process server process cannot be started due to a shortage of kernel semaphores.

   Actions:

   Use the pdsetup command to register the HiRDB directory into the OS.

   Increase the number of semaphores used in the system definition. Note that the new value does not take effect until the OS is restarted.

   Reference:

   This event may also occur immediately after OS startup or when another program registered in inittab is placed in response wait status.