| Yes, it floods those packets with an unknown DA. But only
that first packet; when the station in question ACKs, its
address is then added to the forwarding table.
I don't know the answer to question #2, but it doesn't seem
like a good idea, since that first packet seen for a previously
unknown DA will then be dropped. And you won't be able to
communicate with the poor station that has that address.
Perhaps they wish to explore some type of filtering, if they
are concerned about extraneous traffic traversing the switch
unnecessarily.
-Mike
|
|
I just saw this note, and yes, as Mike said, the DECswitch will
flood the DA packets until they are acknowledged.
The question I have is, does the customer wish to to disable this
behavior because they perceive that they have no control over the
generation of packets with "bogus" destination addresses?
If the customers network consists of a small number of end nodes,
they could put the DECswitch ports in manual mode, add the MAC
addresses of the end nodes that are connected to the specific
DECswitch ports. Thus they give up the learning feature of the
DECswitches, in favor of having a dedicated "hard configured"
LAN that can't evolve on its own, and no bogus or unconfigured addresses
would ever flood to adjacent DECswitch ports. Any future additions of
nodes however, would then have to be manually added/managed (a real pain).
Filtering would minimize flooding but wouldn't eliminate it all
together. There are NIC cards (real cheap ones), that allegedly will
put out extraneous packets with bogus destination addresses.
If this is what the customer is dealing with, then basically they've
gotten what they paid for with those cheap NIC cards.
Bob
|
|
The customer wishes to setup a test procedure to verify that the DECswitch
900EF isolates the various LANs, and one of the tests is to try to loop DECnet
through a bogus address. Using IRIS, I see what appears to be a unicast
packet from one DECnet MAC address to another DECnet MAC address. However,
based on the information that you have provided, the behaviour I am seeing is
the norm and I will just need to document a procedure on how to look into the
packets to determine that they are Connect Initiate to the bogus address, and
the remaining 3 packets are restransmitted CI. I liked doing this procedure
alot better before they added the bogus address!
Thanks for the responses.
Dennis Faust
|