Most important thing about RAID recovery: you need to un-RAID the array and mount each disk as a single disk to the PC where RAID reconstruction is performed. This is an important requirement because it gives access to XOR data blocks and system information about RAID structure which is hidden behind the RAID controller and can not be accessed otherwise.
Please be advised that some RAID controllers will consider the unRAIDed array broken and will no longer recognize it. This happens even if the server was completely powered off before extracting disks. Known RAID controllers that have such feature: MegaRAID, Intel.
Common RAID failures
Usually there are disk issues, some RAID hardware failure, error in system administration (like accidentally format procedure), RAID re-initialization or change of RAID level without backup.
RAID controller problems are easiest to fix. Normally at this case disks are can be read without any problem. All you need to do is to reconstruct RAID using Raid Wizard as some sort of universal RAID controller.
Failed disk isn't tricky. If RAID level stores parity information, all you need to do is to reconstruct the RAID disk in Raid Wizard. Moreover, 80% of such cases are solved by simply starting VMFS Recovery™. It detects and reconstructs the RAID disk at start-up adding both the RAID disk as a physical disk and all the detected volumes from the array.
Let's take a look on recommended approach for all RAID levels:
- RAID 0 – doesn't allows disk failure. If you encounter such a situation the best solution is to get all possible data from a corrupt disk by creating a disk image or using hardware recovery in a specific laboratory.
- RAID 1 – this RAID doesn't need any reconstruction. You can just start reading mirrors one by one as a single drive until you find an undamaged one.
- RAID 5 – allows losing one disk off the set. Please note, you need to specify the original disk number, excluding hot spare drives. Example: RAID5 of 5 disks with failed one disk should be reconstructed as 5 disks set.
- RAID 6 - allows losing two disks off the set. Please note, as for RAID level 5, you need to specify the original disk number, excluding hot spare drives. Example: RAID6 of 8 disks with failed two disks should be reconstructed as a 8 disks set.
- RAID 1+0 – a combination of RAID 1 and 0. In this case, you need to find a working mirror of RAID0. You can use the HEX preview to detect duplicate disks and combine configurations without duplicates.
- RAID 5+0 – a combination of RAID 5 and 0. The methodology is the same as for RAID 1+0. You need to find a working mirror of RAID5. You can use HEX preview to detect duplicate disks and combine configurations without duplicates.
- RAID 6+0 – a combination of RAID 6 and 0. The methodology is the same as for RAID 1+0 and RAID 5+0. You need to find a working mirror of RAID6. You can use the HEX preview to detect duplicate disks and combine configurations without duplicates.
Please avoid adding spare disk into RAID configuration. It will not be detected automatically and will prevent to detection of correct RAID configuration. Moreover, adding an unnecessary disk to Raid Wizard will force it to enumerate the wrong array set and will prevent detecting the correct one.
If you need to reconstruct a 6 disk RAID 5 array where 5 disks are to contain data, and the last one is hot spare, you'll need to choose a 5 disk RAID 5 configuration, leaving a hot spare off the set.