Using database multiplexing mode is the same as obtaining the backup data on the standby server as a safeguard against a disk failure. Note that all server disks may be corrupted due to some cause.
As a safeguard against this type of case, execute the pgx_dmpall command on the primary server to create the backup data.
However, it is not definite as to which server runs as the primary server, so ensure that the pgx_dmpall command is executed periodically on all servers, so that the backup data will be obtained. For example, create a script to obtain the backup data, and set it in the operation management software.
Point
When the pgx_dmpall command is executed on the standby server, it will not match the statuses, however the error message shown below will be output and return the value "1".
If a script that ignores only this type of error is executed on all servers, the backup data of the primary server can be obtained.
Error message
ERROR:recovery is in progress (10095)
Note
Consider the possibility that the server that runs as the primary server may be destroyed alongside the backup data, so it is recommended to promote another server to become the primary server, and then back up the data on the new primary server without waiting for the next scheduled backup.
Specify the same backup directory name for the primary and standby servers. If different backup directory names are specified, and recovery is performed using the backup data of the other server, the recovery cannot be performed correctly.
In the database multiplexing mode, data cannot be recovered to a past point using backup data.
See
Period backups allow shorter recovery time and reduction in disk usage. Refer to "Backing Up the Database" in the Operation Guide for details on the backup operation.
Refer to "Chapter 8 Action Required when an Error Occurs in Database Multiplexing Mode" for details on recovery based on the backup data that was obtained using the pgx_dmpall command.
