Re: Small LC Comment on bfd-base-07

[Jeffrey Haas <[e-mail]>]

On Thu, Feb 14, 2008 at 12:33:51PM -0600, David Ward wrote:
> Vague language is also good if we need to add new/more diags and not  
> be "trapped."

I think this may be less dire than you're implying.

Consider the case where a session not in one of the down states always
resets the Diagnostic to 0 when not in an error state.  In order for the
session to be reset, you must also reset to the down state.  In the mean
time, you've had the opportunity to pull the current Diagnostic off the
wire.  The BFD protocol itself need not necessarily cache this, but the
management infrastructure probably should as part of the "I'm down!"
callback.

Thus, for example, the MIB could have a bfdSessDiagLastDown object along
with the bfdSessDiag object which would represent the current diag value
in the protocol.  Trap objects become slightly more interesting because
they'd might want to represent a transition out of diagnostic state 0
and potentially back.


-- Jeff