2.2.33 Operands related to the system switchover facility

156) pd_ha = use | nouse
Specifies whether the system switchover facility is to be used.
use: Use the system switchover facility.
nouse: Do not use the system switchover facility.
Notes
When Real Time SAN Replication based on the log-only synchronous method is used, the system switchover facility cannot be used at the log application site. When use is specified in this operand, the KFPS04689-W message is displayed.
157) pd_ha_ipaddr_inherit = Y | N
Specifies whether IP addresses are to be inherited when the system switchover facility is used. How the system switchover facility operates depends on whether IP addresses will be inherited; for details, see the HiRDB Version 9 System Operation Guide.
Y: Inherit IP addresses.
N: Do not inherit IP addresses.
Omit this operand when you use the standby-less system switchover facility (this operand will be ignored if specified in this case).
Specification guidelines
If you use the rapid system switchover facility, use the following guidelines when specifying a value:
  • HiRDB single server configuration
    Because IP addresses cannot be inherited when you use the rapid system switchover facility, specify N.
  • HiRDB parallel server configuration
    To inherit IP addresses, specify Y; otherwise, specify N. Note that the units that are the targets of the rapid system switchover facility cannot inherit IP addresses. Therefore, if you specify Y here, you need to specify N for the pd_ha_ipaddr_inherit operand of the unit control information definition of the units that are the targets of the rapid system switchover facility.
    If you specify N for this operand, you cannot specify Y for the pd_ha_ipaddr_inherit operand of the unit control information definition.
158) pd_ha_switch_timeout = Y | N
This operand can be specified when the server mode is used. Specification of this operand is invalid in the monitor mode.
This operand specifies whether to switch the system without waiting for internal termination processing of HiRDB when internal termination processing of HiRDB (a unit for a HiRDB parallel server configuration) during system switchover has exceeded the server failure monitoring time. The server failure monitoring time referred to here is the time specified for the patrol operand of the HA monitor or Hitachi HA Toolkit Extension.
For details about the patrol operand of the HA monitor, see the manual HA Monitor Cluster Software Guide. For details about the patrol operand of the Hitachi HA Toolkit Extension, see the manual Hitachi HA Toolkit.
Y:
Switches the system without waiting for internal termination processing of HiRDB when internal termination processing of HiRDB during system switchover has exceeded the server failure monitoring time. In this case, system switchover is carried out by assuming that HiRDB has slowed down.
If you are using the standby-less system switchover (1:1) facility or the standby-less system switchover (effects distributed) facility, the specification for this operand is invalid during planned system switchover.
N:
Does not switch the system until internal termination processing of HiRDB during system switchover is terminated.
Advantage
If internal termination processing of HiRDB during system switchover takes a long time because, for example, of a disk error, system switchover might be delayed as a result. If you specify Y (default value) for this operand, you can switch the system without waiting for internal termination processing of HiRDB, even when it is taking a long time.
Notes
  • If you specify Y for this operand when the patrol operand value is small, planned system switchover might turn into system switchover based on slow-down. This is because internal termination processing of HiRDB during planned system switchover exceeds the time specified by the patrol operand.
  • You need to be careful if restart is specified for the switchtype operand of the HA monitor or Hitachi HA Toolkit Extension. If pd_ha_switch_timeout=Y (default value) is specified, and if internal termination processing of HiRDB exceeds the server failure monitoring time, HiRDB is not started in the system in which the failure occurred. In this case, the system is immediately switched.
    For details about the switchtype operand of the HA monitor, see the manual HA Monitor Cluster Software Guide. For details about the switchtype operand of the Hitachi HA Toolkit Extension, see the manual Hitachi HA Toolkit.
159) pd_ha_prc_cleanup_check = Y | N
When the server-mode system switchover facility is used, this operand specifies whether system switchover processing is to be placed on hold until all server processes have terminated at the target unit or user server. By specifying Y in this operand, you can guarantee that all server processes that access the shared disk will have terminated before system switchover is performed in response to a server failure.
Note that the switchover target is a unit in the case of the standby system switchover facility or the standby-less system switchover (1:1) facility, and a back-end server in the case of the standby-less system switchover (effects distributed) facility.
Y: Perform system switchover only after all processes have terminated
N: Do not wait for termination of target processes
Before you specify this operand, see Shared disk access control in the HiRDB Version 9 System Operation Guide.
Conditions
All the following conditions must be satisfied:
  • use is used in the pd_ha operand
  • server is specified in the pd_ha_acttype operand
  • The pd_ha_unit operand is omitted
Additionally, if you specify Y in this operand, use an HA monitor whose version is 01-08 or later.
160) pd_ha_mgr_rerun = wait | notwait
This operand is applicable only to a HiRDB parallel server configuration.
Specifies whether to wait for the completion of start processing of other units before switching the system for the system manager unit.
wait:
Waits for the completion of start processing of other units before switching the system for the system manager unit. In this case, the following checks are performed to start the system manager unit:
  • Version check for each unit
  • Is the dictionary server running?
  • Is at least one front-end server running?
  • Is at least one back-end server running?
If another unit has stopped, system switchover for the system manager unit might take a long time or fail. The following table describes what happens when system switchover for the system manager unit occurs when some of the units are stopped.
Reduced activation specification (pd_start_level value)Specification of a name for the unit that does not startAction
Not specified (0)--Cannot start.
Specified (1)Nothing is specified for the pd_start_skip_unit operand.System switchover occurs after a wait time specified by the pd_reduced_check_time operand.#
Stopped unit is specified for the pd_start_skip_unit operand.System switchover is completed immediately.#
Legend:
--: Not applicable
#: System switchover is completed only when all of the following conditions are satisfied:
  • The dictionary server is running.
  • At least one front-end server is running.
  • At least one back-end server is running.
notwait:
Does not wait for the completion of start processing of other units before switching the system for the system manager unit. When this operand is specified, system switchover is executed rapidly for the system manager unit without waiting for other units that have stopped to start.
When system switchover is completed for the system manager unit, the message KFPS05210-I (system initialization completion message) is issued, even if the HiRDB operating environment is not ready. If a job cannot be continued even after the completion of the system manager switchover because of a UAP error, for example, use the pdls command to check the system operating status.
Condition
To specify notwait for this operand, the system configuration must satisfy a certain condition. For details about the required system configuration, see the HiRDB Version 9 System Operation Guide.
Specification guidelines
To prevent failure of the system switchover for the system manager unit, specify notwait. For details about how to handle system manager unit switchover failures, see the HiRDB Version 9 System Operation Guide.
161) pd_ha_transaction = error | queuing
Specifies whether to use the transaction queuing facility. Also specifies the processing that takes place when the number of connections to the HiRDB server exceeds the maximum number of concurrent connections (value specified by the pd_max_users operand) during system switchover. For details about for the transaction queuing facility, see the HiRDB Version 9 System Operation Guide.
error:
  • Does not use the transaction queuing facility. The transactions being processed by the back-end server or dictionary server being switched end in errors.
  • If the number of connections to the HiRDB server exceeds the maximum number of concurrent connections during system switchover, the connections to the HiRDB server end in errors.
queuing:
  • Uses the transaction queuing facility. Instead of ending in errors, the transactions being processed by the back-end server or dictionary server being switched are queued by the front-end server until system switchover is completed. SQL response time will be longer than normal in this case because processing will wait until the back-end server and dictionary server units have started.
  • If the number of connections to the HiRDB server exceeds the maximum number of concurrent connections during system switchover, connection processing to the HiRDB server is retried for the duration of pd_ha_trn_queuing_wait_time + pd_ha_trn_restart_retry_time. However, the HiRDB client version must be 07-00 or later.
Conditions
If you use the transaction queuing facility, all of the following conditions must be satisfied:
  • HiRDB parallel server configuration
  • The rapid system switchover facility, standby-less system switchover (1:1) facility, or standby-less system switchover (effects distributed) facility is used.
Note that these conditions need not be satisfied if connection processing to the HiRDB server is to be retried by a HiRDB client.
Relationship to client environment definition
Even when queuing is specified for this operand, you can specify that the transaction queuing facility not be used for each client. To cancel the transaction queuing facility for each client, specify the PDHATRNQUEUING operand in the client environment definition. For details about the PDHATRNQUEUING operand, see the HiRDB Version 9 UAP Development Guide.
Relationship to other operands
This operand is related to the following operands:
  • pd_ha_trn_queuing_wait_time
  • pd_ha_trn_restart_retry_time
162) pd_ha_trn_queuing_wait_time = transaction-queuing-wait-time
~<unsigned integer>((1-3600))<<180>>(seconds)
Specifies the transaction queuing wait time when the transaction queuing facility is used. If the standby unit or server does not start within the wait time specified by this operand, the transactions being queued end in errors. Transactions that occur subsequently end in errors without being queued. If a unit or server starts before this wait time, transactions are resumed without waiting for the time specified by this operand.
If the standby-less system switchover (1:1) facility is used, the alternate portion becomes the standby system during normal operation, and the normal BES unit becomes a standby system during alternation.
Condition
The pd_ha_transaction operand must be set to queuing.
Specification guidelines
  • Normally, you need not specify this operand. However, if the rollforward during system switchover takes 180 seconds or longer, increase the value of this operand.
  • Specify this value taking into consideration the time required for switchover. When executing a UAP that has a long transaction processing time, the time required for rollforward processing at switchover must also be considered.
  • Choose a value for this operand that reflects the time required for disk switching during switchover. If the specified value is large, it will take longer for the transaction to detect the error if switchover fails. If the value is small, a transaction error might result even when switchover is successful.
Relationship to client environment definition
  • PDCWAITTIME
    If you specify PDCWAITTIME, see PDCWAITTIME in Relationship to client environment definition for the pd_ha_trn_restart_retry_time operand.
163) pd_ha_trn_restart_retry_time = retry-time-upper-limit-after-transaction-start-request-errors
~<unsigned integer>((1-3600))<<60>> (seconds)
If system switchover occurs while the transaction queuing facility is being used, transactions are queued by the front-end server. However, during the period between the system switchover and the restart of the standby unit or server, the front-end server cannot detect the system switchover. During this period (the period between the system switchover and the restart of the standby unit or server), the front-end server issues a transaction start request to the running unit or server. However, because the running unit has already been abnormally terminated, this transaction start request ends in an error. For the transaction that ends in an error, the front-end server re-issues a transaction start request (retries the transaction start request).
This operand specifies the upper limit for this retry time. If the standby unit or server is not restarted within the value specified by this operand, the transactions being retried end in errors. Furthermore, transactions that occur subsequently end in errors without being retried. Note that if the standby unit or server begins to restart before this retry time, no retries occur and transactions are queued.
Condition
The pd_ha_transaction operand must be set to queuing.
Specification guidelines
  • Normally, you need not specify this operand. However, if disk switching by cluster software takes 60 seconds or longer, increase the value of this operand.
  • Choose a value for this operand that reflects the time required for disk switching during switchover. If the specified value is large, it will take longer for the transaction to detect the error if switchover fails. If the value is small, a transaction error might result even when switchover is successful.
Relationship to client environment definition
  • PDCWAITTIME
    Specify the PDCWAITTIME value so that the following relationship is met:
    Value specified for PDCWAITTIME > value specified for pd_ha_trn_queuing_wait_time + value specified for pd_ha_trn_restart_retry_time
    If this relationship is not met, while a transaction is being retried or queued by the transaction queuing facility, the transaction might fail due to the time specified for PDCWAITTIME being exceeded.
    For details about PDCWAITTIME, see the HiRDB Version 9 UAP Development Guide.
Remarks
The relationship between the pd_ha_trn_queuing_wait_time and pd_ha_trn_restart_retry_time operands is explained as follows.
[Figure]
Explanation:
Zones A and D:
Transactions can be started (normal state).
Zone B:
The unit containing the back-end server is being switched, and the front-end server cannot detect this condition. Transaction start request is retried for the duration of time specified by the pd_ha_trn_restart_retry_time operand. When the front-end server detects the system switchover, transactions are queued. If the front-end server cannot detect it within the specified time, the transactions end in errors.
Zone C:
The unit containing the back-end server is being switched, and the front-end server has detected this condition. Transactions are queued for the duration of time specified by the pd_ha_trn_queuing_wait_time operand. If the transactions cannot be started within the specified time, they end in errors.
164) pd_ha_resource_act_wait_time = maximum-wait-time-for-resource-activation
~<unsigned integer>((2-3600))<<10>> (seconds)
This operand specifies the maximum wait time for the running server's resources to be activated at the time of unit startup when you use the standby-less system switchover (effects distributed) facility. Unit startup processing is placed on hold up to the specified amount of wait time. If the resources are activated within the specified amount of time, unit startup processing resumes.
Advantages
When unit startup processing is completed, jobs can be started only if the running server's startup processing is completed in the unit.# By specifying an appropriate value in this operand, you can start your jobs immediately after the unit's startup processing is completed because the unit startup processing will have waited on wait status for resource activation.
#: This event might occur if it takes time to process the command specified in the usrcommand operand in the HA monitor's sysdef file.
Specification guidelines
Normally, there is no need to specify this operand. Specify this operand when all the following conditions are applicable:
  • The standby-less system switchover (effects distributed) facility is being used
  • The KFPS05623-I message was displayed
  • The target unit for the message contains the running server
Use the following guideline to determine the value to be specified in this operand:
10 + time required for resource activation processing (seconds)
Time required for resource activation processing: Maximum processing time for the command specified in the usrcommand operand in the HA monitor's sysdef file
Remarks
If the unit does not contain the running server, the running server's startup is placed on hold for the amount of time specified in this operand. However, if all the servers in the unit start as standby servers, unit startup processing is restarted without waiting for the amount of time specified in this operand.