iSCSI MPIO disconnections

Hello Experts,

We have a test cluster setup with MPIO and all seems to be working well.
The setup is using Windows Server 2012 r2 with a qnap as the backend storage.

Routes are as follows;
172.16.10.10 (qnap port 1)
172.17.10.10 (qnap port 2)

Traffic flows perfectly over each switch with the failover link using MPIO working fine.
The issue we are facing is whenever we disconnect our first link to the qnap to tesy the failover (port 1) the other link kicks in fine however when we restore the 172.16.10.10 link it doesn't automatically map the iSCSI target. So if link 2 (port 2) was to fail the entire cluster fails until we either reconnect the iSCSI target manually or we reboot the server.

My impression is that when a link goes down the other kicks in and once the first link is repaired it should automatically reconnect to the target?

Is there an automatic reconnect setting I'm missing?

Thanks people!
dqnetAsked:
Who is Participating?
 
Brad BouchardConnect With a Mentor Information Systems Security OfficerCommented:
Great question first of all, because there are a ton of people who probably have the same question.  I know I did at one point.

As far as my advice on it, I would have gone down the same path as the user who answered your question on serverfault.com, however the one thing that I wanted to add originally and stopped myself from doing after seeing his 2nd reply was about the Round Robin.  As far as performance goes as long as you have your MPIO targets correctly configured in Hyper-V, or in VMware iSCSI kernels (one per iSCSI target by the way if you ever do VMware), then the hypervisor (Hyper-V in this case) looks for the least used/optimized path/target to send the next connection or set of data over.  I didn't want to sound like I copied him, but Round Robin coupled with the advice he gave you on the active/primary paths makes for some great failover.
0
 
Brad BouchardInformation Systems Security OfficerCommented:
This is ironic, but my search for auto reconnect turned this up:

http://serverfault.com/questions/588177/iscsi-mpio-disconnections

Look familiar?
0
 
dqnetAuthor Commented:
Had to try all my options.. :)

Any opinions on the question though?
0
Simplify Active Directory Administration

Administration of Active Directory does not have to be hard.  Too often what should be a simple task is made more difficult than it needs to be.The solution?  Hyena from SystemTools Software.  With ease-of-use as well as powerful importing and bulk updating capabilities.

 
dqnetAuthor Commented:
Hi Brad,

Thanks for the thumbs up.
I am just a little confused as to what you mean if it is configured correctly in HyperV?
We just use standard Windows Clustering Service.. Is there further setup involved in acheiving MPIO in the HyperV setup after the above?

Thanks mate!
0
 
Brad BouchardInformation Systems Security OfficerCommented:
I am just a little confused as to what you mean if it is configured correctly in HyperV?
We just use standard Windows Clustering Service.. Is there further setup involved in acheiving MPIO in the HyperV setup after the above?
No worries.  Simply put, what I meant was that there were only really two big things to check on and that was if Round Robin was selected and the active/primary target, both of which you already have.  My comment was more for historical purposes so that when someone else reads this post who has the same question/problem as you they will know those are the two key points.

You are golden my friend.
0
 
dqnetAuthor Commented:
Hello again Brad,

Sorry for the delay mate.. I wanted to test it and make sure it all worked first.
And goes what.. It did. What a great setup. It's all switching across seamlessly.
I did have to delete all connections and Luns and rebuilt the entire iSCSI setup but it worked perfectly.

Hope this helps anyone!

Thanks again mate.
0
All Courses

From novice to tech pro — start learning today.