Can restore RAID data by rebuilding RAID structure when data loss occurs with RAID corruption

Problem

Can you restore RAID data by rebuilding the RAID structure when the raid is corrupted and data loss occurs?

There is a large percentage of RAID information loss in a data disaster that occurs in a disk array environment, and many engineers have the experience of rebuilding raid with the original RAID structure and recovering the data. Is this a feasible approach?

Reply

There are roughly several ways to rebuild a raid: Only raid information is created (RAID meta information), only checksums are created when a raid is made (synchronous only), and population initialization when the raid is created.

If the reconstructed structure is not the same as the original structure (involving controller firmware, RAID level, block size, check mode, disk order), the reconstructed RAID LOGICAL driver must be different from the original, so hastily loading the file system will destroy the file system structure and result in data loss.

The following assumes that the reconstructed structure is the same as the original structure:

If the controller's rebuild method is to create only raid information, that depends on whether the previous raid structure is normal or degraded, if it is down-level (already have a hard drive down the line), after the reconstruction, because the data is from the new data and some of the old data combination, the file system will be destroyed, and can not reverse recovery. If the previous RAID structure is intact, rebuilding the raid does not affect the data and can restore the original data exactly as it was.

If the controller is rebuilt by creating a RAID rebuild checksum (even in the background), as in the case above, if the hard drive is offline before, such a rebuild destroys data consistency. If the previous RAID status was online (good), such a rebuild would not affect the data.

If the controller is rebuilt by populating the data (usually 0), it will destroy the data anyway.

In this way, rebuilding the raid after the raid is damaged can restore the data on the premise that the controller is designed to not destroy the data, and that the previous raid state is good, and the structure of the rebuild is exactly the same as it was originally. In addition to the reconstruction there is data risk.

In fact, most raid damage is not from good to paralysis, most of which will pass the downgrade this step, so the actual forced reconstruction of the file system is not a good practice, but the downgrade to the paralysis during the period of time if the data write is not much, Repair file system may be rebuilt after the change that the movie star downgraded to paralysis during this period of time.

Many of the early SCSI-based disk arrays will be rebuilt with at least 0 front-section sectors (such as 1m,10m). At present, the SAS controller based on LSI has a high market share, and its controller often does not purge data when rebuilding, but it is also risky to resynchronize the data in the background.

North Asia Data Recovery Center, http://www.datahf.net