On Sep 5, 2025, at 2:02 PM, Johnny Billquist
<bqt(a)softjar.se> wrote:
Yeah, sorry. I wrote the wrong number.
The problem is still the same, though.
Just because your MicroVAX shows adjacency up don't mean it is so. The other side
(all of them), think that 31.282 is not up. So you only have the circuit up from one side,
which is not enough.
Johnny
Adjacency up involves different checks depending on the node type.
If you're on an endnode, it means simply that a router hello was received (from the
designated router). It does not mean the designated router can hear you
If you're on a router, router adjacency up happens if you hear from the other router
AND that router reports you in its router adjacencies list. In other words, for that
particular case, it does tell you there is two way connectivity.
If you're on a router, endnode adjacency up means you received an endnode hello. It
does not mean the endnode can hear you.
All of the above is for Ethernet; the rules for point to point connections (DDCMP) are
different. Given how the data link layer works there should always be two way
connectivity if any DDCMP traffic gets through at all (if the circuit state is
"on" rather than "on-starting").
The assumption with Ethernet is that one way connectivity is very unusual. Ignoring
newfangled stuff like filtering switches, it only can happen if there is a strange
electrical issue with the cable. A long coax segment with a weak transceiver could give
you this behavior. Similarly, there is an assumption of transitivity: if A can talk to B
and B can talk to C then A can talk to C.
In some limited cases, especially when dealing with router to router connections, DECnet
does not fully rely on these assumptions and verifies them instead. But it does not do
that in all cases, because the issue is too unlikely to worry about in cases where the
network wouldn't utterly fail if it happened.
paul