Solved!
Go to SolutionChanging Drive
Hi Guy's,
We have an Poweredge R910 that saying one of the drives in RAID 5 Failure Predicted YES and it shows Online.
We have bought a replacement Drive and we want to change this drive.
Could you tell us what are the proper steps to replace this driver?
Thanks
1402
0
Posted August 28th, 2018 09:00
Hello
I know I have provided these instructions several times, but I was unable to find a post where I provided them.
- Get a valid tested backup before making any hardware or software changes to a production system
- (Optional) Save a controller/TTY log for your records(Not all controllers support this feature)
- Run a consistency check(Not all controllers support this feature on all RAID levels)
- Offline the predictive failure disk(Not all controllers support this feature). If your controller does not support offlining then reduce I/O on the controller as much as possible and pull the drive.
- Make sure that there are not additional disk failures. If other drives need to be rebuilt then rebuild them before offlining. If offlining the disk will take the virtual disk offline then you will need to perform data recovery.
- Physically replace the disk
- If you insert the replacement disk into the same slot then a rebuild should start automatically. If a rebuild does not start automatically then set the disk as a hot spare to start a rebuild.
- Once the rebuild completes, I suggest running diagnostics on all disks.
You should also review the controller manual for replacement procedures. Also, consistency checks and rebuild operations will affect performance across the controller.
http://www.dell.com/storagecontrollermanuals/
Thanks
1379
0
Posted August 29th, 2018 03:00
Hi Deniel,
This is what I can see regarding the drives and there is not a hot spar in our server.
We dont want any disruption of the data on this server, if I understood you correctly the correct way to replace this drive is to first run a consistency check on the virtual disc, there are 2 virtaul disks in the same RAID 5, should we run the consistency check on both Virtaul disks?
After put the probelmatic drive in offline and then remove the drive and put a new drive in the same slot this way we will not have any disruption, am I correct?
Also you said:
If offlining the disk will take the virtual disk offline then you will need to perform data recovery.
How should we know if offlining the disk will offline the virtual disk?
1321
0
Posted August 31st, 2018 00:00
Hi Daniel,
Do you have an update on my last question?
Daniel My
12 Elder
•
6205 Posts
1305
0
Posted August 31st, 2018 09:00
If you have a sliced array, two virtual disks across the same physical disks, then yes, you should run consistency checks on both of them. A consistency check compares the parity against the data to make sure they match. If there is a mismatch then it attempts to recover the data. You want to do this before replacing the disk because after you replace the disk there will be redundancy for comparison.
The patrol reads that the controller does automatically only check for bad blocks. They do not verify the data. That is the difference between a patrol read and a consistency check.
You should not have any disruption as long as there are no other failed or offline disks in the virtual disk. Performance will be affected during the rebuild and consistency check.
RAID 5 is n+1, it can only sustain one drive failure. Make sure that there are no failed or offline disks. If there are failed or offline disks then you need to resolve that issue before taking the predictive failure drive offline.
Thanks