Status files

Question: Are status files swapped only when a disk error occurs?

Answer

No. Status files are swapped at the following times:

When an error occurs in a status file
When a current status file becomes full
When there is no more usable space in a current status file due to fragmentation
When the pdstsswap command is executed

(2) Does swapping occur even though an error occurs in only one of the status file versions?

Question: Are status files swapped when an error occurs in only one of the file versions?; Does there have to be an error in both file versions before swapping occurs?

Answer

Swapping of status files occurs when an error occurs in either of the file versions. Below is the procedure HiRDB follows from the time an error is detected until swapping is completed:

Searches for normal spare files for both versions A and B.
The contents of the normal system are copied into versions A and B of other status files (the file space is also defragmented).
The status files at the copy destination are used as the current status files.

(3) Handling of status file errors

Question

The following operands are used to specify the HiRDB processing when no more normal status files are available for either file version A or file version B (in the event of an error in file version A or B):

pd_syssts_singleoperation=stop |continue (applicable to unit status files)
pd_sts_singleoperation=stop |continue (applicable to server status files)

Should I specify stop or continue for these operands?

Answer

Specifying stop terminates HiRDB or the unit abnormally. When continue is specified, operation with a single set of status files occurs (this is called single operation).

Status files are important because they contain information needed for full recovery processing. When continue is specified and an error occurs in the status file during single operation, HiRDB or the unit is terminated abnormally because the error is in both file versions. In this case, full recovery processing cannot be executed because neither file version that was used as the current file can be accessed. Following are guidelines for specifying these operands:

When providing for full recovery is more important than preventing abnormal termination of HiRDB, stop should be specified.
When you do not want to stop HiRDB (when you give up on full recovery processing, recover the database up to the point of the most recent backup, and reload data), continue should be specified.

Appendix A.3 Status files

(1) When are status files swapped?

(2) Does swapping occur even though an error occurs in only one of the status file versions?

(3) Handling of status file errors