RE: IPV4LL and EXPIRE
David Hauck
Fri Oct 24 14:39:12 2014
On Friday, October 24, 2014 1:01 AM, Roy Marples wrote:
> On 23/10/2014 22:35, David Hauck wrote:
>>> The scenario is the same regardless of C1/C2.
>>>
>>> Plug cable in.
>>> DHCP starts
>>> If after 5 seconds we have not obtained a working lease, IPv4LL
>>> negotiation will start. The difference is that now the DHCP
>>> negotiation will continue alongside IPv4LL whereas previous it was
> stopped and then restarted.
>>
>> Is there anything different wrt to state sequence? For example, what
>> will
> the state transitions be when IPV4LL is negotiated but then a DHCP
> session is started (is the IPV4LL STOPPED first?)? I'm interested in
> what the DHCP hook state transition sequence will look like.
>
> It may look like this
>
> PREINIT
> CARRIER
> REBOOT
> NOCARRIER
> EXPIRED
> CARRIER
> IPV4LL
> BOUND
>
> When any non IPv4LL address is BOUND, the IPv4LL state is terminated.
> Easier to think of IPv4LL being another DHCP state transition really
> (and this was the case anyway).
>
> The below example this is a valid transition where the DHCP server
> went down and didn't get back up until the DHCP lease had already expired.
>
> BOUND
> RENEW
> REBIND
Shouldn't there be an EXPIRED here?
> IPv4LL
> BOUND
>
> Does this help explain it?
I think so, but: I guess there is no IPV4LL (specific) state transition that indicates the IPV4LL state is no longer active? That is, the IPV4LL->BOUND transition *implies* this but a transition to the BOUND state can now occur from any number of prior states (not exactly sure of the full set of prior states but they at least include REBIND, IPV4LL, CARRIER,...). Perhaps any state transition to (!IPV4LL && !EXPIRED) indicates IPV4LL state machine is no longer active?
Thanks,
-David
> Roy
Archive administrator: postmaster@marples.name