When an error occurs in an audit trail file, the file is placed in shutdown status. An audit trail file in shutdown status cannot be used except to be deleted. Take action according to the following procedure.

Procedure

1. Identify the audit trail file in shutdown status from the KFPS05704-E message.
   You can use the pdls -d aud command to identify the generation in which the error occurred. If HiRDB is running, the auditor executes the pdls -d aud command.
   
2. Use the pdaudrm command to delete the audit trail files in shutdown status. To delete forcibly the files in data load waiting status, specify the -f option.

Files in shutdown status cannot be data loaded. Therefore, if they are not data loaded, the entire audit trail will be lost.

This subsection explains the actions to take if all audit trail files are placed in shutdown status due to a circumstance such as an error that occurs in the HiRDB file system area used for audit trail files.

Processing by HiRDB differs depending on the value specified for the pd_aud_no_standby_file_opr operand.

(a) If down is specified

If HiRDB is active, forcibly terminate HiRDB (or the unit, for a HiRDB parallel server configuration).

If HiRDB is in the middle of startup processing, stop outputting audit trails and then complete HiRDB startup processing.

(b) If forcewrite (the default value) is specified

Stop outputting audit trails. In this case, HiRDB remains active.

The HiRDB administrator must take the following actions.

Procedure

1. If HiRDB is operating, use the pdstop command to terminate HiRDB normally. If normal termination is not possible, use the pdstop -f command to terminate HiRDB forcibly.
2. Use the pdfmkfs command to re-create the file system area for audit trail files. If the HiRDB file system area cannot be used due to a fault such as s disk error, create another HiRDB file system area for audit trail files on a different disk.
3. If you changed the file system area for the audit trail files, also change the specification of the pd_aud_file_name operand.
4. If there is a backup of the file system area for audit trail files, use the pdfrstr command to recover the HiRDB file system area.
5. Use the pdstart command to start HiRDB.

When down is specified for the pd_aud_no_standby_file_opr operand, HiRDB (or the unit, for a HiRDB parallel server configuration) is forcibly terminated if the audit trail files become full or if there are no swappable audit trail files in the event of an error in the current file. The HiRDB administrator and the auditor must take the following actions.

(a) For a HiRDB single server configuration

Select one of the following procedures:

Procedure 1: When pd_aud_max_generation_num < 200

1. Specify 200 for the pd_aud_max_generation_num operand.
2. Use the pdstart command to start HiRDB.
3. Use the pdload command to data load the files in data load waiting status.

Procedure 2: When pd_aud_max_generation_num = 200

1. Specify N in the pd_audit operand.
2. Use the pdstart command to start HiRDB.
3. Use the pdload command to data load the files in data load waiting status.
4. Use the pdaudbegin command to start collection of the audit trail.

(b) For a HiRDB parallel server configuration

Select one of the following procedures:

Procedure 1: When pd_aud_max_generation_num < 200

1. Use the pdstop -f command to terminate HiRDB forcibly.
2. Specify 200 for the pd_aud_max_generation_num operand.
3. Use the pdstart command to start HiRDB.
4. Use the pdload command to data load the files in data load waiting status.

Procedure 2: When pd_aud_max_generation_num = 200

1. Use the pdstop -f command to forcibly terminate HiRDB.
2. Specify N in the pd_audit operand.
3. Use the pdstart command to start HiRDB.
4. Use the pdload command to data load the files in data load waiting status.
5. Use the pdaudbegin command to start collection of the audit trail.

If asynchronous output buffer flush opportunities occur frequently and all sectors of the asynchronous output buffer are placed in flush-wait status, the transaction execution time might increase or HiRDB (a unit in the case of a HiRDB parallel server configuration) might be forcibly terminated, depending on the specification of the pd_aud_no_standby_file_opr operand.

When all sectors of the asynchronous output buffer are placed in flush-wait status, the message KFPS05723-W (when forcewrite is specified for the pd_aud_no_standby_file_opr operand) or KFPS05722-E (when down is specified for the pd_aud_no_standby_file_opr operand) is output. The HiRDB administrator must take the following actions:

• Check the number of audit trails output per unit time, and modify the size and sector count of the asynchronous output buffer if necessary.
  If the size and sector counts of the asynchronous output buffer are insufficient, increase the values specified in the pd_aud_async_buff_size and pd_aud_async_buff_count operands in the HiRDB system definition.
  
• Check the machine and network load.
  If the load is high, lower it. Then restart HiRDB (a unit in the case of a HiRDB parallel server configuration) as needed.
  
• Make sure no error has occurred in the network used for outputting the audit trails from the asynchronous output buffer to the audit trail file.
  If an error has occurred, recover the network from the error. Then restart HiRDB (a unit in the case of a HiRDB parallel server configuration) as needed.