Does the controller recognize it, keeping the data on it? or it starts rebuilding the raid as if it's a new disk?
Asked
Active
Viewed 1,525 times
2 Answers
2
Depends on RAID card firmware. Generally for not crazily old firmware it recognizes the disk as the one already part of the array and does not rebuild it. The point is that metadata is stored on disk with information about array. It is stored on all disks. So if the disk is functional and just hot removed and plugged back in the metadata is intact and it will be compared to metadata on other disks after which it will be just joined into array.
As a matter of fact one undocumented option to temporarily restore failed disk is to hot replug it. Had done that a few times on RAID1 arrays with both disks failed - saves a lot of time and hustle.
Alexey Kamenskiy
- 794
- 1
- 9
- 23
-
Well with PERC raid controllers it is what Dell tells you to do first if a disk fails. Only if it does not go online again further steps are necessary. – Nils Dec 11 '16 at 22:25
-
@Nils thank you for your comment. I have never seen such a recommendation from Dell themselves. I recall seeing this recommendation on some 3rd party blogs not affiliated with Dell. You could help by providing a link to this part of documentation. Though as I note this approach helped me a lot (specially in recovering most of the data from failed disks). – Alexey Kamenskiy Dec 12 '16 at 05:13
-
@Nils Obviously this won't help to recover all data as the disk failure in PERC controller is triggered by certain count of media errors, which means some blocks with data are actually dead. – Alexey Kamenskiy Dec 12 '16 at 05:14
1
Generally - yes, the raid controller will detect the topology changed twice and should start the rebuilding process. Raid metadata will also indicate that the raid content has changed (if it has changed).
drookie
- 8,625
- 1
- 19
- 29