RE: [Rstplib-users] RE: Root port waits for 'slow' Designated Port in .1w ?

[Mick S. <mic...@ie...>]

The Visio will let you see all the state variables on each port, so you can
construct a scenario and match an implementation step by step against it.

-----Original Message-----
From: own...@ma...
[mailto:own...@ma...]On Behalf Of Alex Ruzin
Sent: Tuesday, November 27, 2001 1:50 AM
To: mic...@ie...; std...@ie...;
rst...@li...
Subject: RE: [Rstplib-users] RE: Root port waits for 'slow' Designated Port
in .1w ?



Dear Mr. Seaman,

On Tuesday, November 27, 2001 9:48 AM you wrote:

Mick> I really do suggest you get your management to buy you a PC
Mick> and a copy of Visio so you can test this and all the other cases
Mick> you will find while debugging your code with a tool (the simulation
Mick> avaible on the .1 website)
Mick> that a significant number of others have already debugged.

I have just checked the Visio. It helps to see, *what must* be,
but doesn't help to understand *why* & *how* and why my (or any
other) implementation doesn't work properly. It could help here if the
sources is open, but they are not :(
You don't want to public the sources as 'procedural model',
like 8.9 in 802.1d, do you ?

Anyway, I will use Visio to verify my implementation.

From other side I tried to make my state machines implementation
as formal as possible. And my questions (I hope so) are related to
the .1w, not to the implementation and bugs in it.
[For example, the result of our previous discussion "indirect link
fail in 802.1w" was, that I implemented some your important change
in the official edition of .1w]

> It is true that a Designated Port with a point to point link
> to an Alternate Port does not transition to Forwarding
> quickly. To do so would violate some
> of the (current at least) assumptions that underlie the proof
> that Root Ports can be transitioned without permission from
> their neighbour without causing loops. Fortunately since
> the Designated Port only leads to the
> Alternate Port and to no (other) end stations there is no
> denial of service.
>
> It is not clear from your description where the "problem" of
> non-communication is. However I would suggest that rrWhile
> should be cleared as soon as all prior root ports are no
> longer listening or learning (and that this is already specified in
> the standard)

Would you like to point me to the place in the .1w ?

> ...so there is little (read as short as an individual
> bridge  implementation can manage it)
> delay for C2 to
> become Forwarding once it is Root Port. You seem to be
> missing the code that causes ports that have rrWhile not
> cleared to revert to Listening once
> reRoot is set by another port.

You are completely right, in the case port  C1 has to hop to
DESIGNATED LISTEN and all will be OK!
So, after this your mail it works fine.

Thank you very much, Alex