Recovery of an online table reorganization

The failure of an online table reorganization is often due to processing errors, such as disk full or logging errors. If an online table reorganization fails, an SQLCA message is written to the history file.

If a failure occurs during run time, the online table reorg operation is paused and then rolled back during crash recovery. You can subsequently resume the reorg operation by specifying the RESUME option on the REORG TABLE command. Because the process is fully logged, online table reorganization is guaranteed to be recoverable.

Under some circumstances, an online table reorg operation might exceed the limit that is set by the value of the num_log_span database configuration parameter. In this case, the database manager will force the reorg utility and put it into PAUSE state. In snapshot monitor output, the state of the reorg utility will appear as PAUSED.

The online table reorg pause is interrupt-driven, which means that it can be triggered either by a user (using the PAUSE option on the REORG TABLE command, or the FORCE APPLICATION command) or by the database manager in certain circumstances; for example, in the event of a system crash.

If one or more database partitions in a partitioned database environment encounters an error, the SQLCODE that is returned will be the one from the first database partition that reports an error.