Hi Sandy,
I just replied to Russ's message before I read your response.
The draft language is just as Russ said.
[MZ] That's OK :)
It is important that the VLAN membership of all the RBridge ports in
an LAALP MUST be the same. Any inconsistencies in VLAN membership
may result in packet loss or non-shortest paths.
Following this, there is a paragraph fully describing an inconsistent
configuration.
[MZ] Yes, that paragraph further explains the possible consequences of
inconsistency.
If it is not possible for inconsistencies to occur, then why are you
mentioning
them? Why give an explicit example of an inconsistent VLAN configuration?
[MZ] This is to explain why consistency becomes a requirement for LAALP
implementations that are connected to TRILL edge. Implementations that cannot
fulfill this requirement should not be used to connect to TRILL edge. IOW, it
is not qualified as an LAALP as defined in TRILL. You can also refer to the
second 'c)' of Section 2.1 [RFC7379].
You say "if the consistence cannot be maintained, it is not qualified as a
LAALP".
But the example shows a configuration case where inconsistency happens.
[MZ] As explained above.
Does that mean it is not an LAALP?
[MZ] No, it's not an LAALP as TRILL defined and required.
Does the spec for LAALP ensure
consistency (and if so, why mention inconsistencies and provide examples)?
[MZ] There are several proprietary link aggregation techniques in the industry.
It's possible that some of them does not meet this requirement. So the
requirement is a criteria for choosing LAALP implementations. I think this is
easy for operators.
It's possible that operators want to do manual configuration. In that case, the
consistency requirement still holds.
Does anything in TRILL detect the absence of consistency?
[MZ] I think the detection of inconsistency is out the scope of TRILL. However,
it's natural that an edge RBridge device implements both TRILL and LAALP. This
device can detect the consistence by using LAALP.
Thanks,
Mingui
-Sandy