Case:
A server of the client room IBMX365M3, 4 146GSAS hard disks, and 4 pieces formed a RAID5. The No. 3 plate has already been offline. After that, the 2nd plate is offline, and RAID collapses.The operating system is WindownS2003SP2, and the application system is a framework on Oracle and CAD data. The user requires that the database and CAD images are recovered as much as possible.
Solution:
Restore plan: 1. Protect the original environment and close the server to ensure that the server is no longer turned on during the recovery process. 2. Sign the fault hard disk to ensure that it can be fully restored after taking out the slot. 3. Mount the fault hard disk to the only environment, and make a complete mirror on all the fault hard disks. After the backup is completed, the original fault disk is returned, and the subsequent recovery operation will not involve the original fault disk before the data is confirmed. 4. An analysis of the RAID structure of the backup disk, obtain its original RAID level, band rules, strip size, verification direction, META area, etc. 5. Create a set of virtual RAID5 environments based on the obtained RAID information. 6. Check the virtual disk and file system. 7. Detect whether the virtual structure is correct. If it is incorrect, repeat the 4-7 process. 8. After determining the data is correct, the data is relocated according to the user’s requirements. If the original disk is still used, it is necessary to determine that after the original disk has been completely backup, the RAID is rebuilt, and then relocated. When relocating the operating system, you can use WinPE and other to carry out. You can also use another hard disk to install a relocation operating system on the fault server, and then the sector -level relocation. Recovery process: 1. Use Winhex to make the original hard disk full mirror and make 4 146G mirror files. 2. Analysis structure: The best structure obtained is 0,1,2,3 sequences, lack of 3 plates, block size 512 sectors, BackwardParity (adaptec). 3. Data verification after the group, the latest compressed package of more than 200m or more has no error, and the structure is correct. 4. Directly copy the database file and CAD file. 5. The database engineer resumes and restores the database according to the copied database file. At this point, the entire data recovered successfully.
