The home port along with the home node of a LIF is a property of the LIF. There is no reason to recreate it to change it. By design, the home port and home node can be modified at any time with the command
substituting in the correct values for the SVM (vServer), LIF, node, and port as needed. If the node remains the same, you can just change the home-port. When you run the above command, nothing happens to the LIF as viewed from the outside world. The LIF stays working on the same port, in your case e0M. But, it is not marked as being on a "non-home" port. You can then use the command
which will cause the indicated LIF to migrate to it's home node and port from wherever it is currently located.
It is important that you verify network connectivity and setup on the new home port are correct. That is the e0M physical port and the a0a-25 VLAN-ed ifgrp should exist in the same broadcast domain, VLAN, etc. Otherwise while the LIF might move, it could interrupt communication. Assuming that your network setup is correct, any clients connected through the LIF as it exists on e0M will just switch over there communication to port a0a-25 without any noticeable delay.
An alternative set of commands to accomplish the same task would be to use
which will move the LIFs port location first. But, that one does not update the home port definition - you'll still need to run the "network interface modify" command as above. The choice is one of style. You can set the home location then move, or move then set the home location. It just takes two steps.
To recreate the LIF you'd have to take the LIF offline for a short while as you'd be deleting and recreating. Using either of the two modify/move sequences will keep the port online and let normal network protocol mechanisms discover the port change automatically.
When the system was removed from the network, it wasn't shutdown gracefully and network (1 and 10G) were literally ripped out of switch and filer. And best of all the documentation was destroyed. I've been working this for a couple days. I researched the NetApp text book install and configure pages to insure the ifgrp, vlan and LIFs were correct as well as ports on the Nexus 3200 switch. My best guess is that when I reconnected the nodes to the switch, I cause the nodes to failover to a LIF that didn't exist.
I did the migrate command, but got the following error:
Error: command failed: LIF "a0a-2" on vServer "filer_fs01" could not be found in database (entry doesn't exist)
1. Is this cDOT or 7-mode? Based on the original post and discussion of LIFs and home-node I assumed cDot, as those are cDot style terms, but with the information you added I have enough to question and I just want to confirm.
2. Can you relay the command you tried to use for the migrate? The reason I ask is that I think you've confused some information. The error message implies that you used the "port" name of a0a-25 as the name of a LIF on vServer filer_fs01 as opposed to the LIF name.
Key information : Ports (such as a0a-25) are associated with nodes (technically a type of SVM/vServer, but not really the same as a "data" SVM vServer). LIFs are associated with user created "data" SVMs. Again generalizing, as "intercluster LIFs" for replication are associated to the "cluster" SVM, which again technically an SVM but not a "data" SVM. It gets a bit confusing if you're not familiar with it.
Output that would be good to have:
set -rows 0
network port show
network interface show
The first command turns off the pagination on output. The second lists the ports by node. The third lists the LIFs by SVM. This way we can be sure ports are defined where expected and LIFs are known as we check out the exact form of the commands.
Of course you could also make it be home on node phoenix-02 instead if that's more appropriate.
I see that LIF vi01_mgmt1 on SVM phoenix_vi01 isn't at it's home port. If port a0a-25 is where you want that one to run normally, you should consider updating the home node and port for that LIF as well.