SeeDk
asked on
RAID 1 disks sliced into multiple Virtual Disks..is this a good configuration?
This is on a Dell Powervault MD1200 which is serving as storage for an ESXI5.5 host.
There are two 8TB disks for this purpose.
They will act as storage for an RHEL VM server.
Had problems getting it working when setting up as one large VD occupying the entire available space: Link
Thought it was the disks - got new ones: same issue.
Decided to create a VD using only 2TB...success! It works as intended. Seems to be some bug in the Linux or ESXI OS.
I am creating more VDs with the remaining space to make sure this will work.
What I am wondering if this is a good idea? Any potential drawbacks? Is it harder to recover from a disk failure or other such problem?
There are two 8TB disks for this purpose.
They will act as storage for an RHEL VM server.
Had problems getting it working when setting up as one large VD occupying the entire available space: Link
Thought it was the disks - got new ones: same issue.
Decided to create a VD using only 2TB...success! It works as intended. Seems to be some bug in the Linux or ESXI OS.
I am creating more VDs with the remaining space to make sure this will work.
What I am wondering if this is a good idea? Any potential drawbacks? Is it harder to recover from a disk failure or other such problem?
ASKER
Issue was random lost partitions, data loss. Completely unstable. There's more details in that topic I linked but essentially it was not possible to create a stable partition on Linux.
I agree Linux is the issue since I tested on a Windows VM and didn't see this problem. However, I ran the latest RHEL release and still the problem remained.
Don't need high IOPS for this - it is just storage.
VMDK sizes I started with 4TB and tested recreating vmdk's all the way down to 450GB - seeing the issue at every size.
VMFS5 is used on all datastores.
I agree Linux is the issue since I tested on a Windows VM and didn't see this problem. However, I ran the latest RHEL release and still the problem remained.
Don't need high IOPS for this - it is just storage.
VMDK sizes I started with 4TB and tested recreating vmdk's all the way down to 450GB - seeing the issue at every size.
VMFS5 is used on all datastores.
So if there was no issues with Windows, the problem definitely lies with Linux distro.
as for your question, no issues with creating a RAID 1 Datastore for VMs.
What model are you using here, MD1200, because I'm not see-ing it listed in the HCL ?
as for your question, no issues with creating a RAID 1 Datastore for VMs.
What model are you using here, MD1200, because I'm not see-ing it listed in the HCL ?
ASKER
The PowerVault MD1200 is a DAS.
The ESXI server running on a PowerEdge R730
The ESXI server running on a PowerEdge R730
Okay, not seeing the MD1200 DAS on the HCL.
So if the storage is not on the HCL.....
Most MD1200 are connected to a MD3200, I've seen not direct to Host.
Have a look and see if you can find it on the HCL.
So if the storage is not on the HCL.....
Most MD1200 are connected to a MD3200, I've seen not direct to Host.
Have a look and see if you can find it on the HCL.
ASKER
Can't find it either. Guess that means it's not surprising to run into a bug like this?
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Thanks for the help!
If you only have two disks, there is not much you can do with them for a RAID configuration.
More Disks, would give your more IOPS.
I assume you were creating a VMFS datastore to store the virtual machine disk, so it does not matter how large a VD you create on the MD1200.
The HOST VMFS datastore presented to the ESXi host, has got nothing to do with the virtual machine disk, or VM OS.
e.g. the VM OS cannot see the VMFS datastore, it can only see the size inside the VMDK you create.
So Linux OS is the issue, based on the size you create your VMDK, not the size of the LUN on the MD1200.
What size virtual machine disk are you creating ? 8TB, 2TB ?