NETROM Oddity

[N5X%PSUVM.BITNET@Sds.Sdsc.Edu (814)]

>I netrom conn'd to a node, it only took 3 retries to get to him (two
>hops away), but as soon as he accepted the connection my end told me
>it had timed out? Thereafter, the other end continued to send me CONN
>ACKs which my end sent back. This continued with the ttl going down
>until the packet expired.

>I later received a netrom Info frame with the BBS login (the other
>station was also running NOS kh113014), which again I ignored (it was
>RR'd at ax25).

>Q: Shouldn't netrom at least deal with netrom traffic destined for my
>own node?

This could be related to a bug in the netrom code I ran into a while back.
The time required to for a connect ack to come back after a connect request
is sent is timed in     nr4.c for use as the irtt for the connection.  This is
done in an 16 bit integer in millisec,  so if it takes more than 32 sec
to connect there is overflow.  This results in a bogus 'timeout' response
which kills the connection at the origionating end only. Note: if there is a
transport layer retry before the connect act arrives, then the default irtt
is used and the problem doesnt show up.

There doesnt seem to be anything like a RST packet in the netrom transport
protocol, so the answer to your last question is that NetRom ignores any
transport packets that arrive for a connection to its own node that it doesnt
have a record of.

73's Jim KB3KJ