home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Ham Radio 2000
/
Ham Radio 2000.iso
/
ham2000
/
tcp_ip
/
rose_nos
/
rosehdr
< prev
next >
Wrap
Text File
|
1992-02-26
|
5KB
|
88 lines
ROSE Circuits -- The Header
When you connect using a ROSE network, you issue the connect to the
destination via your neighbor ROSE's call and the destination ROSE
switch's address. e.g. C DEST V MYROSE-5 405999, where 405999 is the
address of a ROSE that is near DEST. (These examples will use the TAPR
commands. With NET you leave out the V and add the interface name in
front of the destination call sign. Also, these examples will not
include the optional (one each) digipeater calls between you and your
neighbor ROSE, or between the destination ROSE and the destination.)
The neighbor rose switch is watching packet frames fly by and in those
that have two to four digipeaters in them it looks for its call in the
first digipeater call with a clear "has been repeated" bit, followed by
a ROSE address of the right format. If it finds one, it intercepts your
SABM, sending a UA and an I frame back to you as though they had come
from DEST. If all goes well, the ROSE switch on the far end correspond-
ing to 405999 address sends an SABM to DEST while assuming your call in
its header source field. (All call signs in the header's digipeater
fields have their "has been repeated" bits set.) As observed by DEST,
the headers for this and all following frames to DEST show your call in
the source call sign field and his neighbor ROSE switch's call in the
last (or optionally, next to last if there is an "exit digipeater")
digipeater address field. (The ROSE address number that preceeds the
ROSE call sign in the header is the address number of the ROSE switch on
the other end of a circuit, but that is not significant to us except in
setting our arp entry.)
Here is an enquiry to "USERS" on a ROSE switch which is sustaining one
of our IP circuits. Irrelevant lines have been deleted.
K5JB-11 AX25L2 User Linked to USERS @ 3100405732
K5JB-10 *** Unknown Protocol ID CC ***
K5JB-11 *** Unknown Protocol ID CC ***
N0ELS-5 X.25 Trunk (R1) with the following connections:
K5JB-10 @ 3100405771 ( 1 P4 D1) --> K5JB-11 @ 3100405732
In this example, K5JB-10 is an IP station with N0ELS-5 as a neighbor
ROSE (address 405771). K5JB-11 is an IP station with K5JB-5 as a neigh-
bor ROSE (address 504732). This USERS report was received from K5JB-5
(405732) by initiating an AX.25 connect from K5JB-11.
There is some difficulty in switching routes around while using IP over
ROSE because it gets easily confused with our antics. If you are going
to try another ROSE switch after making configuration changes it may be
best to make an AX.25 connection to the destination IP station though
the ROSE circuit, disconnect it, and make sure you don't have any lin-
gering AX.25 connections to the ROSE switch before trying a new scheme.
This method may not work until the ROSE switch breaks down its internal
network state. The only way to force this _may_ be to create a second
circuit using a different routing. ROSE seems to delete the old circuit
when you do that. Some of the early problems I had were caused by my
impulsiveness. I haven't had enough experience to tell yet but I sus-
pect that once you establish a virtual circuit over ROSE it stays intact
unless you take some action to disconnect it.
(Not sure when to stop rambling about T3, AX.25 version 1/2, etc. Some
of these things should be in the general knowledge of AX.25 users but I
am sure they are not.)
ROSE "Reliable" mode
If you use the ROSE switch's digipeater call instead of its primary call
in making a ROSE connection you will encounter the "reliable" mode. It
does cut down the spurious frames from the ROSE switch, but it becomes a
rather fragile circuit that won't allow a reset. If it works for your
circuit, use it, but if you have trouble setting up a circuit after
restarting NET, use the regular ROSE call instead.
On the other hand, maybe the circuit breakdown with a reset is just what
we need to be able to try various circuits without waiting for the ROSE
switches to retire their table entries.
Other Kinds of Circuits
As an afterthought, this scheme might be needed to make IP work over
TEXNET. I don't have a neighbor TEXNET to play with, but I could have
even used my old "Gator" switch with this method. Also, if the IP PID
in transmitted headers causes a problem with the terminus of someone's
circuit scheme I can replace it with no layer 3 PID and nobody would
know the difference.
If you want to play with the source code for the rose trap you can
certainly have it. I will add it to the TAPR release kit after it has
been thoroughly wrung out.
Joe, K5JB