08-20-05 07:46 AM
darek> Could somebody explain me the mechanism/procedure of loop
darek> master selection in Fibre Channel Arbitrared Loops? I
darek> wonder wchich device becomes the loop master - is it the
darek> one with the lowest WWN?
This is true, except when a switch port is on the loop (a public
loop). In this case, the switch port always becomes loop master.
I'm not sure if you wanted more detail about the LISM process than
that. Roughly, here's how it works:
1. When a NL-port comes online (or begins to re-initialize in
response to a LIP), it begins continuously emitting LISM
frames containing its own port wwn.
2. Each NL-port listens for LISM frames. When it receives a
LISM,
(a) if the wwn in the incoming LISM is less than the
wwn of the receiving port, then the receiving port
stops sending its own LISM frames, and begins to
pass received LISM frames through unmodified.
(b) if the wwn in the incoming LISM is greater than
wwn of the receiving port, the receiving NL-port
drops the incoming frame and continues emitting
its own LISM frames.
(c) if the wwn in the incoming LISM is to the wwn of
the receiving NL-port, then this LISM frame has
been passed on by all other NL-ports on the loop,
and the receiving NL-port is the loop master.
(I don't have the relevant fibre channel specs to hand, so I've
probably got some of the details wrong. If you really need to
understand this stuff, go to www.t11.org and start downloading the
latest public versions of the specs. Be prepared to spend a lot of
time reading.)
Note that LISM frames don't really have a destination per se; they're
received by the next NL-port on the loop and processed as described
above. In the description above, I've omitted the special cases
necessary to force an FL-port, if any is present, to be elected loop
master.
Dale.
[ Post a follow-up to this message ]
|