Notifications
Clear all
Managing Disks - Certify, Verify, Initialize
4
Posts
2
Users
0
Reactions
11.1 K
Views
Topic starter
I have a "failure predicted" message for one of the drives in my RAID 0 volume. Is it possible to copy the data off the disk being replaced to the new disk in order to not have to erase the whole volume.
To put it another way, I want to leave all the data on disks 1-3, copy the data off disk 4 to the new disk 4 that is replacing it, and then start the volume up in the same way as it was previously. Is this possible? If so, what is the recommended method?
Posted : 24/06/2016 3:10 pm
It is possible, but not recommended.
The only way is to use the command line dd command to copy the entire disk, then swap them. If you get it wrong, you lose all data, so we do not encourage this.
Posted : 24/06/2016 4:06 pm
Topic starter
Thanks for the info. I actually didn't get a notification via email that there was a reply to my thread—hence the delay.
To be safe, I just copied all data to another disk and then set up a new RAID volume and copied it all back. But I appreciate the info—it may be useful to me in the future.
If I did use this method, would I know immediately if it didn't work properly? Could the volume start up and I could then notice file corruption down the line? Or is this a case of the files on the volume not even being recognizable if it didn't work? I ask because it may be worth a try in the future as long as I have a backup, but I would only want to do this if there would be an immediate indication of any problem. In other words, I wouldn't want to find out a year down the line that data was corrupted.
Posted : 03/07/2016 4:46 pm
No, if you do this correctly, you will immediately have a good volume. (or no volume at all!)
The only cavaet is if there is parts of the disk which are damaged. then the terminal dd command will just "skip" those bad sectors, meaning the data is misaligned on the disk. There is a option in dd to avoid this. However, dd does not tell you that data was unreadable, and when a portion of a disk is unreadable, the option fills it with zeros on the duplicate. Bottom line is yes you could have bad data on a copied disk that you do not discover for some time. So in the case of a failing disk, duplicating the failing disk in a stripe to a new one has that extra risk.
Posted : 04/07/2016 1:12 am
