Recovery Server Error Handling
New options are provided for handling errors during recovery redo and undo processing.
Offline redo/undo processing occurs during installation startup if the installation previously shut down abnormally. Before other servers start, the Recovery Server processes the log file. It first re-does all updates in the log from the last consistency point, then un-does any open transactions.
Online undo processing occurs when the Recovery Server is performing pass-abort of transaction (after a rollback failure by a DBMS Server).
Previously, if an error occurred during this processing, recovery of the database was stopped, the database was marked inconsistent, and the Recovery Server startup was halted. The only option left to the DBA in such a situation was to roll forward the database. This feature provides additional options for handling such errors.
Two new Recovery Server parameters in CBF allow additional options for handling Recovery Server redo/undo errors:
• offline_error_action
• online_error_action
Note: You should use the new configuration parameters instead of the unsupported environment variable II_DMFRCP_STOP_ON_INCONS_DB. If II_DMFRCP_STOP_ON_INCONS_DB is set, the configuration parameters are ignored.
For more information about these configuration options, see the following sections in the Database Administrator Guide:
• Recovery Server Offline Error Handling
• Recovery Server Online Error Handling