When running a Restore, the log states the number of records read for every restored object. The log specifies the number of records that were successfully restored, how many failed, and how many were skipped.
When retrying a Restore job, the number of successful cases and failed cases may not add up to the number of failed cases in the initial Restore job.
This is because the retry will only attempt to restore failed cases that were caused by 4 ”retryable” errors:
Any failed records caused by another type of error are not retried, and not included in the final count of the retried Restore job.
In the following example, in the initial Restore, 434464 cases failed. In the following Restore attempt, 419693 succeeded, while 14762 continued to fail. We can assume that 9 cases were not "retryable" errors, and therefore were not attempted in the second Restore.