RAID 5 - 4x drives. 1x drive failed. Another classed as missing !??!
Hey All,
Urgent question for anyone please.
I have got a RAID 5 config on a Data volume in a Dell server. One drive went pop today and has failed.. still spins up (as in can feel when ejected) but is classed as failed. This should be ok as my other drives should still run. Oddly another drive is classed as MISSING in the raid controller. Although when I inspect it, it is set as FOREIGN and the RAIDC sees its capacity.
Now I have pulled the bad one and reinserted thats classed as MISSING.
I have tried reinserting the curious MISSING drive.. but no joy..
Original failed drive blinks orange.
MISSING drive is Green. No signs of failure.
I have lost the volume in Windows server 2012 R2... Any ideas as the restore will take some some some time to another server.
Would like to get the volume up and replace the bad drive and let it rebuild but VD is offline due to it seeing 2 drives off.
Any ideas given tonight would be wonderfully accepted :)
OS still boots as the OS VD is good.
Picture of RAID controller screen attached.
Many thanks in advance
R
IMG_20160824_182928_resized_20160824.jpg
Urgent question for anyone please.
I have got a RAID 5 config on a Data volume in a Dell server. One drive went pop today and has failed.. still spins up (as in can feel when ejected) but is classed as failed. This should be ok as my other drives should still run. Oddly another drive is classed as MISSING in the raid controller. Although when I inspect it, it is set as FOREIGN and the RAIDC sees its capacity.
Now I have pulled the bad one and reinserted thats classed as MISSING.
I have tried reinserting the curious MISSING drive.. but no joy..
Original failed drive blinks orange.
MISSING drive is Green. No signs of failure.
I have lost the volume in Windows server 2012 R2... Any ideas as the restore will take some some some time to another server.
Would like to get the volume up and replace the bad drive and let it rebuild but VD is offline due to it seeing 2 drives off.
Any ideas given tonight would be wonderfully accepted :)
OS still boots as the OS VD is good.
Picture of RAID controller screen attached.
Many thanks in advance
R
IMG_20160824_182928_resized_20160824.jpg
ASKER
I could swap the corrupt drive out... but I dont think it would rebuild as the VD is offline as it looks like 2 drives are missing.
I could try to import foreign config.... the drive at some point might have been dislodged and has been booted without... maybe,....
Whats peoples opinions please.
thanks
R
I could try to import foreign config.... the drive at some point might have been dislodged and has been booted without... maybe,....
Whats peoples opinions please.
thanks
R
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
I would generally agree with rindi's comments. RAID 5 is generally to be avoided....
Nonetheless, if you would otherwise give up on the old drives, I'd give a shot at importing the foreign configuration. One possible scenario that got you here was that the "Missing" drive was disconnected at some point and wasn't noticed. When the other drive failed, the array was no longer usable.
The problem is that if you bring the "Missing" drive back in, it will have data on it from whenever it went offline. Files written since that time won't be recoverable as you'll have newer data from the 2 good drives and older data from the previously "Missing" drive. This might get you some older data files safely, but that's about it.
The fact that this is a VD will likely complicate matters.
On the other hand.... if the drive went "Missing" in the very near past, you may have a shot.
Restoring from a recent backup is likely the most expeditious way to get this running properly.
Nonetheless, if you would otherwise give up on the old drives, I'd give a shot at importing the foreign configuration. One possible scenario that got you here was that the "Missing" drive was disconnected at some point and wasn't noticed. When the other drive failed, the array was no longer usable.
The problem is that if you bring the "Missing" drive back in, it will have data on it from whenever it went offline. Files written since that time won't be recoverable as you'll have newer data from the 2 good drives and older data from the previously "Missing" drive. This might get you some older data files safely, but that's about it.
The fact that this is a VD will likely complicate matters.
On the other hand.... if the drive went "Missing" in the very near past, you may have a shot.
Restoring from a recent backup is likely the most expeditious way to get this running properly.
ASKER
HI guys.
I have restored some data already to an alternate location.
Restoring the rest as we speak but running out of disk space.
I do now have 4x good disks. I dont think I need to do a RAID recovery etc... as my backup looks good. Just wanted to get it up and running and start swapping disks to get the array 100%.
So I guess I can either
a) Put a new disk... DONT think it will build tho as we are missing a disk... and its RAID 5
b) I try to import the foreign config and ... 'see what happens'. If all dies.. then swap the disks out and restore back to a new VD.
If I try A... whats your thoughts on this effecting me running B ??
Cheers guys
R
I have restored some data already to an alternate location.
Restoring the rest as we speak but running out of disk space.
I do now have 4x good disks. I dont think I need to do a RAID recovery etc... as my backup looks good. Just wanted to get it up and running and start swapping disks to get the array 100%.
So I guess I can either
a) Put a new disk... DONT think it will build tho as we are missing a disk... and its RAID 5
b) I try to import the foreign config and ... 'see what happens'. If all dies.. then swap the disks out and restore back to a new VD.
If I try A... whats your thoughts on this effecting me running B ??
Cheers guys
R
ASKER
And the question is... views on RAID 10 .. apart from losing lots of disk space?
R
R
RAID 10 is a lot better than RAID 5, and it mainly gives you a speed increase.
Just kill your current array and build a new one with RAID 6 (maybe get one or more additional disks if you need the space).
Just kill your current array and build a new one with RAID 6 (maybe get one or more additional disks if you need the space).
ASKER
I know the speed increase is V good.. But the 4x drives is the max for that server... so I cant make RAID 6 as the array was pretty darn full to start TBH... however, might time for a spring clean.
Thanks
R
Thanks
R
It looks like you have just come across the RAID5 problem of not noticing the first disk has failed until the second one fails, at that point you are dependant on your Backups!
RAID5 is now not recomended for Business use due to the HIGH risk of a second disk failing while the first failure rebuilds
RAID10 is the way to go, disks are cheap, your data isnt!
RAID5 is now not recomended for Business use due to the HIGH risk of a second disk failing while the first failure rebuilds
RAID10 is the way to go, disks are cheap, your data isnt!
ASKER
Can I get a recommendation on the following please.
Going to install two new servers to replace what I have and I am virtualising everything.
End situation will be two identical servers (arriving tomorrow :) ) and probably 3-4 VMs on one server than will either make a cluster or Hyper V replication.
So in this situation, rather than losing 50% of my storage by using RAID 10... I know there is great read and write gains, should I possibly use RAID6 and only lose two drives with only READ gains.
If theres a total disaster the second server will either kick in or be spun up?
This might be a new question... but its brought the plans forward due to that drive array issue.
Any feedback?
Thanks everyone
R
Going to install two new servers to replace what I have and I am virtualising everything.
End situation will be two identical servers (arriving tomorrow :) ) and probably 3-4 VMs on one server than will either make a cluster or Hyper V replication.
So in this situation, rather than losing 50% of my storage by using RAID 10... I know there is great read and write gains, should I possibly use RAID6 and only lose two drives with only READ gains.
If theres a total disaster the second server will either kick in or be spun up?
This might be a new question... but its brought the plans forward due to that drive array issue.
Any feedback?
Thanks everyone
R
You could also use several RAID 1 arrays rather than a large RAID 10 array. You can then split your VM's among the arrays, that way the disks won't be a bottleneck if all the VM's are running. Using OS software RAID for those arrays you get better read speeds as part of the data is read from one disk, while the other part from the other disk. Hardware RAID controllers will generally not be able to do that as efficiently.
RAID 6 on the other hand would be needed if you really need a lot of storage space.
RAID 6 on the other hand would be needed if you really need a lot of storage space.
ASKER
Thanks Rindi.
That is the amount of storage I will get with the RAID 10... but no gains on the write.
But in my VM environment, do you think this would still be a better throughput option.
Due to budget the OS and the Data will be on the same VD in a Raid 10 or possibly I can give them a RAID 1 if I use the RAID 6. This would give me a redundant OS boot of 300GB and data VD of 2.4TB. Which will work for me and into the future.
Just thought RAID 10 might be a way to go forward... but I lose a lot of disk.
Thanks
R
That is the amount of storage I will get with the RAID 10... but no gains on the write.
But in my VM environment, do you think this would still be a better throughput option.
Due to budget the OS and the Data will be on the same VD in a Raid 10 or possibly I can give them a RAID 1 if I use the RAID 6. This would give me a redundant OS boot of 300GB and data VD of 2.4TB. Which will work for me and into the future.
Just thought RAID 10 might be a way to go forward... but I lose a lot of disk.
Thanks
R
I would prefer using several RAID 1 arrays rather than a single RAID 10 array, for the reason I mentioned above, the load is better balanced among the VM's.
Rather than partitioning your first array for the OS and DATA, I would suggest splitting that array into two volumes. Most RAID controllers allow you to do that. The first volume would be small for the OS, the other large for the data. That way, within diskmanagement you will see two separate disks, rather than a large one with partitions. It is easier to manage disks than partitions.
Rather than partitioning your first array for the OS and DATA, I would suggest splitting that array into two volumes. Most RAID controllers allow you to do that. The first volume would be small for the OS, the other large for the data. That way, within diskmanagement you will see two separate disks, rather than a large one with partitions. It is easier to manage disks than partitions.
ASKER
Thanks Rindi.
I think I misguided you.
I meant more
RAID Controller level:
Array A - Raid 1 - OS HOST boot.
Array B - Raid 6 - OS HOST Data
VMs then live on the Array B which is one large volume in Windows.
Didnt quite follow what you mentioned. I might have explained badly. I was not talking about paritioning the first array. I meant only Two arrays in total for the server. A for boot and B for working/VMs/Data.
Thanks
R
I think I misguided you.
I meant more
RAID Controller level:
Array A - Raid 1 - OS HOST boot.
Array B - Raid 6 - OS HOST Data
VMs then live on the Array B which is one large volume in Windows.
Didnt quite follow what you mentioned. I might have explained badly. I was not talking about paritioning the first array. I meant only Two arrays in total for the server. A for boot and B for working/VMs/Data.
Thanks
R
I would use:
Array 1 = RAID 1
Array 2 = RAID 1
Array 3 = RAID 1
...
On Array 1 you can setup 2 Volumes. Array 1 Volume 1 and Array 1 Volume 2.
Array 1 Volume 1 would hold your host (it can be kept small, the bare-metal Hyper-V will probably have enough space with 50GB).
Array 1 Volume 2 would be the datastore for your first VM.
Since these Volumes are defined in your RAID controller, they appear as separate disks to the OS, even if they share the same RAID array.
Array 2 would be the datastore for your 2nd VM.
Array 3 would be the datastore for your 3nd VM.
and so on.
Of course this assumes you can live with the space you get on your different RAID 1 arrays (you could also use different sizes of disks that accommodate for the different tasks of your VM's).
Array 1 = RAID 1
Array 2 = RAID 1
Array 3 = RAID 1
...
On Array 1 you can setup 2 Volumes. Array 1 Volume 1 and Array 1 Volume 2.
Array 1 Volume 1 would hold your host (it can be kept small, the bare-metal Hyper-V will probably have enough space with 50GB).
Array 1 Volume 2 would be the datastore for your first VM.
Since these Volumes are defined in your RAID controller, they appear as separate disks to the OS, even if they share the same RAID array.
Array 2 would be the datastore for your 2nd VM.
Array 3 would be the datastore for your 3nd VM.
and so on.
Of course this assumes you can live with the space you get on your different RAID 1 arrays (you could also use different sizes of disks that accommodate for the different tasks of your VM's).
ASKER
This was the best course of action.
Thanks everyone. Foreign did not work. Array was toast.
I suspected as much however, restores were taking sooooo long.
Thanks everyone for your help
Thanks everyone. Foreign did not work. Array was toast.
I suspected as much however, restores were taking sooooo long.
Thanks everyone for your help
ASKER
When I checked the volume was gone.
Was copying lots of data across network at the time AWAY from the machine from a remote server
And also a hard drive was plugged in.. All of this pretty much within the 10 minute window of it happening.
Have not inspected the logs yet in windows as I wanted to get the thing off as no data is a no good server for this one and jump into the Raid Controller at boot.
R