Case:
IBM V7000 storage is mounted on a WINDOWS2003 server. The storage space of 7TB is only divided into one partition, and the drive letter F is allocated. One day, the client restarts the server after the server is suddenly DOWN. The disk is scanned and repaired, but the repair process is still not completed after 12 hours. At this time, the customer forcibly shuts down the server and disconnects the connection between the server and the storage. After starting the operating system and then connecting to the storage, the drive letter of the F disk can appear, but When opening it, it prompts “The file or directory is corrupt and cannot be read”, check the partition properties and become RAW format.
Solution:
Step 1: In order to ensure 100%of the data of the data, the original data backup of the source storage was first made before the recovery. Contact the hardware manufacturer engineer, and use the IBM V7000 curl image function to use the optical fiber switch to the IBM DS4800 storage through the optical fiber switch. All operations followed by backup. Step 2: The underlying analysis of the mirror volume is found to be analyzed, and the failure point is the 80 attribute DataRuns operating system of the MFT table itself. The number of files is very huge reaching more than 21.4 million files, the MFT file itself has reached more than 20 g, and the MFT file has a large amount of fragments. The third step is to use two ways to perform data recovery operations: A. Use data recovery software scanning Data in the partition, but I tried the two data recovery software for a full scan. The scanning results were not ideal. There were more than 6T data. You can only see more than 2T data, and the directory structure is chaotic. It is estimated that the number of files is too large. B. Use manual analysis to collect all MFT table fragment information, and rewrite the 80 attribute DataRuns of MFT based on the collection of fragments. This method is very time -consuming, but the effect is very good. Open the partition through the software, you can see more than 6T data, and the directory structure is complete. Step 4: Export data, first spending one night, exported more than 2T Oracle database files, and the software company verified that the database recovery was perfect; the remaining files were exported, and the remaining files were fragmented. The export speed of small files is very slow, and it takes 8 days to export the remaining data for 8 days.