HPE Storage Users Group

Steve,

We are confused that these arrays cannot be purchased without support.

Moreover, I've been working with hp long enough to know that even if you didn't have support, their advice is free. They won't send you a new node, but they'll help.

The main issue here though isn't that you won't call hp. Maybe your in iran and you can't. The main issue is you've (again) made some very poor decisions, and approached the array in a hasty, ham fisted manner.

Regardless of the support status, your approach needs to change, and you'll prevent a lot of these issues

For your own sake, you need to learn to treat this equipment with respect.

As far as your issue, its really pretty cut and dry. You are getting a remote node mismatch error. What that means is node 3 is expecting to see say node 1 at the other end of the cluster link, but its seeing node 2.

It's a human friendly error. Unless there are some other errors you haven't shared, thus can only be caused by the nodes being in the wrong place

Hi Schmoog
I talked with HP. they want to charge us for new controller but I think the problem it's not from the controller . be honestly I fixed the first problem in node 2.before when I got a console of node 2 I saw a node 1 but now it's working and when I got a console of node 2 I saw node 2.I think the problem is from The IO in some our shelf. I need more time. I'll let you know as soon as possible my friend. I'm in Armenia now.

Steve,

I'm a big believer in that the reason for an error message is to tell you what is wrong.

Therefore, if the error says "remote node mismatch", then the problem is that the remote node is mismatched. Plain and simple. Unless there's a different error message, I'm not seeing anything for you to base the belief that the problem stems from shelf IO on.

Anyway, HP's advice is not without merit. The reason is that while you didn't complete the initialization before this error was made, you initialized it enough for the nodes to pair with each other. Replacing the node with a brand new node will allow the new node to pair correctly. It's a bit of a lazy way to fix the issue IMO based on the evidence that you've presented before us, but it would get the job done.

What we really need is the error message from each node, labeled by node position. So for instance:

Node 0 position - <ERROR MESSAGE HERE>
Node 1 position - <ERROR MESSAGE HERE>
Node 2 position - <ERROR MESSAGE HERE>
Node 3 position - <ERROR MESSAGE HERE>

Hello Guys
Finally I fixed the problem ))))))))))))))))))))))))))) .the problem was at node 2.
Schmoog , thanks a world for your useful advice.

King Regards

Hello Steve Carlose,

please tell us, how you fixed this issue.

Thanks and best regards
Sebastian

Hi wagnerse
I unset the cluster in node 2 with unset cnt_no_cluster=2 command change the HBA and one IO again set the cluster in the node.

Regards

That's certainly one way to do it.

So rather than relocate the node to the correct slot, remove the cluster from the node, and allow the node to assume the identity of the slot in which it's located.

Are there any data in this array ?
If not then uninitialize the array, connect to node 0 with serial cable, and restart the MOB.