Formulating the state machine in the reference draft, I come to the conclusion that we should actually *not* mention that "reset to default configuration" in it, because it actually has nothing to do with AN at all.

From an AN perspective it´s important to show what happens when:
- we delete the LDevID. A factory reset does that.
- a device loses all ACP tunnels.

The configuration of the device is outside scope, I think! If a config change removes the last ACP tunnel (for example), then it's that removal that changes the state in the AN machine.

The important thing to mention (and so far missing) is IMO that *only* a factory re-set can remove the LDevID, otherwise we may end up with stale state at enrollment time. I added that point in my new section.

Please shout if you think I'm missing something.

Michael


On 11/03/2017 21:00, Michael Richardson wrote:
Michael H. Behringer <[email protected]> wrote:
     > worth noting in the reference draft though, to be sure).  - A process
     > where the LDevID remains on the device in my view of the world is
     > therefore NOT a factory reset. I would call this "erase device
     > configuration except the LDevID".

"Reset to default configuration"

     > I therefore suggest to use / define the term "factory reset" as per
     > first bullet above. And NOT define two types of factory reset. It just
     > feels wrong to me.

I agree.

     > What am I missing? Why did we even need a term for the second? Can we
     > not just say "delete config, but leave LDevID"?

I think that this is the activity that one usually wants when some
button is held down.  Deleting the LDevID is probably more than most
operators expect.

Deleting the LDevID brings the device back to "unowned" state.
Some devices should support such a thing, and some simply should not.
Some should require a screwdriver/jumper/etc.

In particular, being able to reset to default configuration may be
something that should be easy to do for a untrusted operator
(i.e. CPE in residential setting), but which shouldn't change the onwership.

  --
Michael Richardson <[email protected]>, Sandelman Software Works
  -= IPv6 IoT consulting =-




_______________________________________________
Anima mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/anima

Reply via email to