Dear Den,

Thanks for these useful comments on the PCEP FSM.

I'm bundling the answers:

Comment 1:

> In the description of the Idle state you can find:
> 
> " - Initiates of a TCP connection with the PCEP peer,
>   - Increments the TCPRetry variable
> "
> 
> 1)This seems to be in contradiction with the definition of 
> the TCPRetry variable.
> 
> 2)This variable can be only incremented in TCPPending state, 
> because only in that state it is possible to determine if 
> connection was established or not. 

You are right, the processing is not consistent with the definition.
This will be updated in next revision.

Comment 2:

>The reception of an malformed message will be mentioned in OpenWait 
>and UP state, but not in KeepWait state. Why?

This is an omission that will be rectified. Of course this applies to
this state as well.

Comment 3:

> In the description of this state:
> "
> 2.  If the proposed values are acceptable, the system adjusts its
>        PCEP session characteristics according to the proposed values
>        received in the PCErr message restarts the KeepWait timer and
>        sends a new Open message.  If RemoteOK=1, the system 
> stays in the
>        KeepWait state.  If RemoteOK=0, the system moves to 
> the OpenWait
>        state.
> "
> 
> If moving to OpenWait state OpenWait timer should be started 
> and not Keepwait timer. In the OpenWait state it will not be 
> waited for the expiration of the Keepwait timer.

Actually the OpenWait timer has been started earlier. Here you send an
Open, so you start a Keepwait timer. 
But I agree there is something missing. We need to process
Keepwait/OpenWait timer expiration events both in Keepwait and OpenWait
states. 

By the way that would actually be simpler to define a single timer to
control session setup, as in BGP.

Note that the FSM will be consolidated in rev 08, and this will account
for your comments.

Kind Regards,

J-L

_______________________________________________
Pce mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/pce

Reply via email to