Ok, WOW that is a mess.
Someone at one point in time felt it was needed to give two Physical adapters, two VIO adapters, and then somewhere they built a failover/link aggeration adapter with what I am going to assume is built from the two VIO Ethernet's



So ent adapters are used to specify the hardware adapter. Has nothing to do with TCP/IP address.
en represents the interface associated with hardware adapter. So, en0 *should* reference ent0. I say should because you can do "stupid" things like renaming an adapter!
et is used for IEE 802.3


Likely you are seeing an IP alias, that's why you have two ip's on the same adapter.

Now, as to two adapters with same IP, one could be disconnected at the physical layer.
**EDIT again. You can assign an IP to the ETherChannel adapter, and then the parent adapter. But its not a good way to go about it. That actually looks like what is happening here. Or someone tried to remove a backup adapter from the EtherChannel and gave it the IP.

To be honest, I'm not even sure where to begin with cleaning this up, or how far you want to go.
I'd recommend tracking down what ip's and what adapters are actually in use.
For the physical adapters, entstat -d <entX> should return in the output a 'PHYS_LINK_UP'
I am unsure on tracking it from the VIO side of things. You would have to trace from the client lpar, to the vio and look at the configuration.


From what you posted, it really looks to me like someone took the VIO ethernet adapters (ent4 and ent6) to create the EtherChannel adapter ent5, and en5 is your actual link to the outside world.
If that's the case, that's not an ideal configuration.
You should build your EtherChannel adapters at the VIO level, so ent1 and ent2 to make ent3, present ent3 up to the virtual switch so the client uses ent3 from the vio's for communication (client lpar would see it as ent0). Even if you had dual lpars, the setup would be the same because the virtual switch would handle any failover between the vio's.


There's a very good chance to clean this up properly you may have to sell an extended downtime.


**Edit: Also concerns me that the VIO may not be properly configured as well.