Re: [pim] Use of BS_Min_Interval in BSR draft

[Alexander Gall <gall at switch.ch>]

On Mon, 22 Jan 2007 10:19:12 -0800, Rishabh Parekh <rishabh at ipinfusion.com> said:

> On Monday 22 January 2007 07:14, Alexander Gall wrote:
>> Throttling is not necessary in a stable network because a candidate
>> BSR does not generate any BSMs as long as it receives those from the
>> elected BSR (i.e. this particular event does not occur at all in this
>> case).  Thrashing during BSR (re-)election is prevented by
>> BS_Rand_Override, so this issue appears to be moot.
>> 
> The BS_Rand_Override interval at a new C-BSR might not be large enough (and 
> thereby contained within the 60 second BS_Period window at E-BSR) to prevent 
> the C-BSR from transitioning to E-BSR from P-BSR state and sending out an 
> empty BSM. In this case, the E-BSR will send out a BSM (if the new C-BSR is 
> inferior) without waiting for BS_Min_Interval.

Yes, this can happen, but it has no effect on the receiving routers
because the RP-set doesn't change.  I really don't think it matters.

> Even if throttling is not required in above case, I feel that the MUST 
> statement in the spec should be modified to explicitly exclude "No Forward" 
> BSMs and Unicast BSMs from BS_Min_Interval limit to prevent a 
> misunderstanding as evident in earlier mail from Sumanta.

Note that such BSMs are originated by the DR.  So this is only
relevant if it happens to be the E-BSR.  IMHO, a clarification for
this corner case is not needed.

--
Alex




_______________________________________________
pim mailing list
pim at ietf.org
https://www1.ietf.org/mailman/listinfo/pim