niculelu
1 Nickel

hi,

in solaris i see that a disk is failed.so i replace it in vmax 40k.After i replace the disk physically should i execute further things in SAN like recreate LUN or others.the pool is raid5.tnx.

Labels (1)
Tags (1)
0 Kudos
1 Solution

Accepted Solutions
Jasonc1
3 Argentium

Re: disk

Jump to solution

You are correct in that a single disk failure in RAID5 is rebuilt from the other members in the raid group - no actions needed at the lun level.

VMAX systems contain hot spares, so the drive rebuild is automatically performed to the hot spare typically. This all occurs transparently to the host environment.

Should there be multiple drives fail in the same RG, you should contact EMC support immediately. The array will dial home anyway provided there is connectivity and a maintenance contract in place. It may be possible for EMC support to rebuild the raid group without going to backup, but yes, it is quite possible that you will face a data-loss situation and require a recovery from backup.

Note that in an SRDF/S environment where the VMAX is replicating to a remote array, the remote array acts as an additional mirror, so less chance of data loss in a compromised RAID group scenario.

0 Kudos
8 Replies
Jasonc1
3 Argentium

Re: disk

Jump to solution

Can you add a little more detail here please? You are saying that the Solaris OS noticed that one of it's 40K luns was in a failed state is that correct? It is not going to occur because of a single drive failure. Multiple failures that affect the integrity of the raid group / pool will be noticed by the OS. In this case you should have EMC support check it out.

0 Kudos
niculelu
1 Nickel

Re: disk

Jump to solution

hi,

I manage solaris and crate LUN is SAN.

In solaris when a disk failed it can be seen with commands.Next i need to replace this disk.In every replacement prcedure after i unconfigured the disk in OS i physically replace the disk in SAN.

My question is -after i physically replace the disk in SAN there is require to perform other things in SAN?

tnx

0 Kudos
Highlighted
Jasonc1
3 Argentium

Re: disk

Jump to solution

I'm sorry, I am not getting what you mean here.

So you are using Solutions Enabler on your Solaris host to work out if a disk has failed or the OS is seeing a failed drive?

The 40K presents a device out to the world and it is supported on the back-end of the array by the physical disks that are installed and the raid / pool configuration. A single failed drive that is raid protected is not visible to the host OS. This is why I am having trouble with your question.

Replacing a failed disk drive is transparent to the hosts connected to the array. Provided multiple drives in the same RAID group have not failed. If there are multiple failed drives in the same RG then you are in a data unavailability or data loss situation. A simple drive replacement will not get the environment back to health here.

niculelu
1 Nickel

Re: disk

Jump to solution

hi,

You are right.There is a raid in SAN so a disk failed is not seen by OS.

The procedures i seen shows the disk replacement from server itself.

Let say in SAN is a raid 5 and after i replace a disk then the data is created on the new disk from parity.right?So no other things should be done like LUN creation.If 2 disks fails in raid 5 then OS will see the disk failed and the raid group must be recreated in SAN and restore from backup.ringht?

tnx.

0 Kudos
Jasonc1
3 Argentium

Re: disk

Jump to solution

You are correct in that a single disk failure in RAID5 is rebuilt from the other members in the raid group - no actions needed at the lun level.

VMAX systems contain hot spares, so the drive rebuild is automatically performed to the hot spare typically. This all occurs transparently to the host environment.

Should there be multiple drives fail in the same RG, you should contact EMC support immediately. The array will dial home anyway provided there is connectivity and a maintenance contract in place. It may be possible for EMC support to rebuild the raid group without going to backup, but yes, it is quite possible that you will face a data-loss situation and require a recovery from backup.

Note that in an SRDF/S environment where the VMAX is replicating to a remote array, the remote array acts as an additional mirror, so less chance of data loss in a compromised RAID group scenario.

0 Kudos
niculelu
1 Nickel

Re: disk

Jump to solution

hi,

just to clarify.

when i create a lun in RG this lun can be mapped to any hba port from SAN controllers.right?tnx

0 Kudos
KW160
2 Iron

Re: disk

Jump to solution

Correct.

0 Kudos
niculelu
1 Nickel

Re: disk

Jump to solution

tnx

0 Kudos