Upgrade Firmware and storage disconnect

Fred Hubbs · ‎04-17-2012

During the firmware upgrade to version 2.0.2q everything went well until I upgraded the 6124 Interconnects. I first upgraded the secondary then the primary. During both upgrades the ESX Hosts reported loss of access to the datastore attached to the hosts. The guests continued to ping which leads me to believe that the host did not really loose access because the guests run from the datastore reported as lost in vCenter. Within 8 mins or so the storage error corrected. My question, is this normal behavior. Has anyone else seen this behavior during interconnect upgrades?

Joseph Ristaino · ‎04-20-2012

Hey Fred,

You have multipathing software installed correct? If so then the failover should have been seamless.

Did vCenter report a loss of redundancy or a loss of access?

If you run the "show service-profile circuit server x/y" command from the fabric interconnect during these I would be interested to see if the other fabric interconnects vHBA remained up.

If multipathing software is installed on the host and 2 paths to the LUN are presented then failover should have been seamless.

The datastore that is lost was the boot LUN as well?

Joey

Fred Hubbs · ‎04-20-2012

Hi Joey:

Thank you for your reply. Yes multi pathing is in place and failover does occur. This was my test environment I did first. I just noticed that of three hosts two reported redundancy link lost (which is normal) but one showed redundancy link and the managment network lost and access to the datastore for VM guests although VM guests continuded to ping normally. The host however was not pinging in fact as being reported.

Anyway, I went forward with production last night and all went as expected with only redundancy link lost during interconnect updates/reboots. No problems. So, I believe my test environment has a connection between the two interconnects switched around from when I first set it all up. There have been some others working in my test area.

Thanks again,

Fred