There is no need to stop the I/O on these hosts .Same is meant by Brendon and jbcool1.
Keep a copy of the old shelf firmware handy in case you need to roll back.
In worst case if the ESH module goes unresponsive,still your disks will be available on the second path (I am assuming that all your shelves are multipathed).Call up the NetAPp support immediately then.
You have said v38 so I am guessing you mean FC disks rather than SATA. I have upgraded many (120+) shelves in the last couple of years and never had any issues. You did not say that you had a HA pair or multipath so you will need to reboot the filer in order to upgrade the firmware.
It has been my exprience that the firmware upgrade takes 10 minutes per shelf. Sometimes it can be faster if you are lucky. The process works the way described by NetApp and is straight forward.
I have never created extra backups because of firmware upgrades but we do have a 2nd snapmirror destination site, so the risk is much lower than a single filer environment.
If you log into NetApp my autosupport you can get a health report for your filer. Also think about an BIOS, disk firmware, RLM upgrades. The health report will let you know what needs to be upgraded.
I've just checked all my devices for BIOS, Shelf and Disk FW this afternoon. As it goes mine was all up to date but i'd go along with Brendon. I don't take additional backups before i do this. I don't have HA systems but i do have my shelves multi-pathed. This means that shelf FW updates can be done without reboots.
the Netapp instructions supplied with the FW updates are very easy to follow.