Start a Conversation

Unsolved

This post is more than 5 years old

219032

January 12th, 2016 12:00

MD3820i MPIO

I am running a MD3820i (storage) with 2 controllers. I have 4 nodes (server 2012 r2) configured and connected to the storage however it seems that only one path is being used.

I read in a post that this is because only one(1) controller is the preferred path for the LUNs and the other is fail over. If that is true how can i get both controllers to be active so the load can be shared.

Other suggestions as to why i am having only one path being used are also welcomed.

January 13th, 2016 09:00

Hello, storagesupport.

How many virtual disks do you have on the storage? These controllers are definitely 'active/active'. If you have more than 1 virtual disk, I would recommend  redistributing the virtual disks across both controllers.

This is how you would assure each controller "owns" an even amount of work, and ensure even load balancing between the two. 

Beyond that, I would verify your settings in the iSCSI initiator, and assure that you're up to 'best practices' for these nic's. You can find a decent walk through in the administrator's guide (page 61) here: http://downloads.dell.com/Manuals/Common/powervault-md3400_Reference%20Guide8_en-us.pdf

Here's the Deployment guide as well: http://downloads.dell.com/Manuals/all-products/esuprt_ser_stor_net/esuprt_powervault/powervault-md3820i_Deployment%20Guide_en-us.pdf

It has a good walk through as well for iSCSI setup.(page 41)

I hope this helps. Let me know if you have any questions.

January 13th, 2016 12:00

yes its only 1 virtual disk ... and all the iSCSI initiator is setup correctly.

January 13th, 2016 14:00

So, if there's only 1 virtual disk, then you'll see it owned by only 1 RAID Controller. It will not change ownership unless in the state of failure.

Now, if you're seeing only 1 path used in your initiator, that's a different story. However, this 1 RAID Controller ownership of the Virtual Disk is by design.

January 14th, 2016 11:00

OK that's good to know that its by design. So when i said 1 path I meant it to be 1 controller, both IP are active to that controller per node.. so I guess based on your explanation its working fine by design.

If that controller failed would the system automatically switch to the second controller or do i need to make it the preferred owner of the disk?

can I add more drives to the storage and expand the VD without loosing data.

Thanks

January 14th, 2016 14:00

Good questions!

So, the design is: IF something along the path fails, whether it be from the cables to the RAID Controller, the other RAID Controller will take the data request. It will also throw an error in the form of an alert in your MDSM that says, "Virtual Disk is Not On Preferred Path". This message is to let you know, that something failed, and the other RC is handling the IO.

Then you can deal with that as necessary.

Adding disk space to the VD is a simple (but tedious) process via MDSM. This can be done without losing data.

While the disks are being added, you could say it's vulnerable in that the parity is being recalculated, however, the task is not destructive.

I wouldn't recommend adding space to the virtual disk during peak hours, but at the end of the day / slow time / etc.

I hope that helps.

No Events found!

Top