This post is more than 5 years old

197 Posts

1769

May 6th, 2009 05:00

Changing FCID for AIX Host implications

We are planning on moving several AIX machines to a new set of switches. This is going to cause the FCIDs to change for each AIX port. If I power down all the hosts and then make the change will I still have access to the my LUNs when bringing the hosts back online with different FCIDs?

30 Posts

May 6th, 2009 16:00

Provided that both your AIX version & HBA types (+ firmware) are reasonably current you should be able to set the `dyntrk` attribute on the fscsiX device.

More details are at http://publib.boulder.ibm.com/infocenter/systems/index.jsp?topic=/com.ibm.aix.prftungd/doc/prftungd/dynamic_tracking.htm

6 Operator

 • 

1.5K Posts

May 6th, 2009 08:00

So you are replacing the existing Fiber Channel switches by new ones. In that case, as I understand, there should not be any issue accessing the LUNs on the hosts when they are back on the new switches. You are not changing anything on the hosts or on the array.

However, others may correct me if needed and provide valuable suggestions.
Thanks,
Sandip

2 Intern

 • 

1.3K Posts

May 6th, 2009 09:00

I guess you wil have two paths at the least; So move only one at a time so the chnage would be trasparent to host with the help of Path management s/w.

6 Operator

 • 

2.1K Posts

May 7th, 2009 07:00

Some recent experience on this has led me to believe that you shouldn't run into a problem if you are just moving host connections on the fabric. We have only ever run into problem when we move the array connections. I'm not sure why this makes a difference, but it seems to.

So, as already suggested you should consider moving just one redundant connection at a time and then rescan to pick up the new paths. Once everything is stable again you can move the other connection and then scan for those new paths.

If you get through the first stage on a few servers and see no impact of just moving the host connection point to the fabric, then you can safely proceed without having to worry about the rescanning. I would still go for just one connection at a time though so you can leave your hosts up and running the whole time.

197 Posts

May 14th, 2009 09:00

Thanks everyone for your responses. The first few servers that we completed worked out well. The server team decided they wanted to power them down and move the cables at that point. All of them came up fine with access to the their disk.

If in the future we have a hard time getting the downtime we would probably just go the 1 port at time approach to handle it.

Thanks again!
No Events found!

Top