Start a Conversation

Unsolved

This post is more than 5 years old

1322

March 8th, 2010 01:00

McData DS24M2 firmware upgrade

Hi

I have 2 McData DS24M2 attached to the DMX1000. The current firmware is at 6.02.00.

I would like to upgrade the firmware to 9.6.

On these 2 SAN switches, we have AIX, Solaris and HP-UX 11i.

As the servers are mission-critical, I would like to do it the fabric firmware  online upgrade, one switch at a time.

For AIX, Solaris and Windows, we have PowerPath. For Powerpath, we can set the HBA to maintenance mode.

For HP-UX, we are using pvlinks. As we are on MCSG, if we disable 1 HBA, will it cause a failover?

Thanks

1 Rookie

 • 

20.4K Posts

March 8th, 2010 02:00

it should not cause failover, pvlinks should redirect i/o to alternate links defined in your volume group. Make sure all pvlink settings are correct (disk time-out ..etc).

2.1K Posts

March 8th, 2010 11:00

You shouldn't have to do anything special on the host except to make sure that the dual pathing is all working just in case.

I have done many firmware upgrades on DS24M2 switches and never experienced any connectivity outages during the upgrade. I believe that the warnings in the documentation are simply there as a CYA. That being said, I assume you are going to be upgrading in stages... right? Personally I would recommend planning to hit every major release along the way until you finally get up to 9.6

The other question is, do you have a specific reason for stopping at 9.6? Unless I'm mistaken 9.9 is available as a final state for these switches.

127 Posts

March 8th, 2010 16:00

Hi Allen,

As per my EMC CE, he wants HP to confirm that by putting 1 HBA to maintenance mode (using fcmsutil disable)

it won't cause a failover. I have a 4-node MCSG cluster.

Do I have to set the pvlinks disk time-out to 90 as per my EMC CE recommendation. These HP servers are running

Oracle DB.

Thanks

2.1K Posts

March 9th, 2010 08:00

I guess if your CE is insisting on it then you will have to proceed that way. We have never seen the need for it, and have upgraded switches with both HP and MS clusters active on them.

No Events found!

Top