Help settling a debate about loopback-detection!

[Cristek]

Hi guys, there’s a bit of an internal debate here in the office about this feature!

Loopback-detection, do you use it on edge/host ports only, or, do you use it on every port, even uplinks?
(assume STP is on and working as intended)
And whatever your preference, can you elaborate on why?

[frank]

Anywhere where STP is disabled
On edge switches where there is a high turn over of devices where end user might connect 2 ports together (lab, conf rooms)
On redundant network paths where STP is disabled

[silvio]

I prefere it to use at ALL access ports. On top (like frank has written) at all ports without STP (mostly to switches from other vendors). Also at the ports to unmanaged switches.
If you use it at uplinks between switches, than the uplink will be blocked and a lot of devices behind the access switch are isolated.

[Cristek]

And when does it make sense to use remote-origin?
None of us here can come up with a scenario where remote-origin solves a problem that STP wouldn't catch anyway, so can anyone provide an example or some input on this?

[silvio]

STP is not working in a lot of cases f.e. at UNP-Ports, DHL, SPB, some stupid switches/phones
Remote-origin is necessary if the loop is between different omniswitches (not within the same VC). Sometimes in one floor you have two VC. And if someone make a connection between them and STP is not working....
There is no problem (beside to write two lines in the config) with remote-origin. But in case of that scenario this can be very helpfull.