[Dxspider-support] Add IP address tp PC16

[Dirk Koopman G1TLH]

Lee Sawkins wrote:
> The most important use for DX Clusters is to distribute DX spots.  Yet
> these DX spots account for a minor amount of the traffic.  The PC92
> traffic is many times the DX spot traffic.  This is mainly the "who is
> connected" traffic.  If you could turn off the PC92 traffic for all your
> node links except one or two reliable ones, you would greatly reduce
> your backbone traffic and not lose anything.  Better yet, only use a
> couple of node links.

My take on this subject is probably not as simple as Lee's. From where I
sit, I see only a very moderate increase in abuse (whether in spots or 
other avenues), certainly compared to previous abusive events that we 
have seen on here.

I also think about the larger picture of the network and how various
bits of protocol traffic affect that. As far as PC92 goes, I tend to
agree with Lee that the product of the number of nodes, their
interconnections and the number of users, is now too much to continue to
track user connections / disconnections. This does not mean that we
should throw away all PC92 traffic because, again from where I sit,
there is useful information in there.

I propose that I modify the DXSpider defaults to disable PC92 A, D for
users (only) and C traffic completely. I wish a) to retain PC92 A & D
for nodes b) PC92 K for keepalive and general statistical purposes and
c) add IP addresses to node PC92 A records both for diagnostic purposes
and future enhancements.

This will mean talk messages will always be flood routed, much as
announces and chat messages currently are.

I will be adding a command to obtain a PC92 C record on demand from
another node - again for diagnostic purposes.

I don't know who sends personal messages (as opposed to bulletins) via
the cluster system these days but one of the consequences of disabling
user PC92 A & D is that users won't, automatically, get sent their msgs
when they login. In order to get over this, I shall probably have to
implement the PC92 F or P command that "finds" or "pings" for a callsign 
by flood routing so that if a personal message needs to be sent a route 
can be discovered. For efficiency, that route will need to be 
remembered, when the reply comes back, by the intervening nodes. 
Assuming, of course, that the recipient has not chatted or spotted or 
announced first (i.e. the normal case :-()


> 
> I would really prefer to only see the IP addresses in DX spots and not
> anywhere else.  Since the number of these spots is small, the increase
> in data would be too be small.

I have no problem seeing PC61s with IP addresses attached. It may prove 
a useful diagnostic tool. I will probably standardise on it myself.

> 
> I am quite willing to share my IP to country list with Spider.  This
> list is quite accurate.  Today with this list I dropped all the DX spots
> from W0RLD and GL0BE, which were all Spam about global warming.  The
> last dropped spot was from the call E51XIW complaining about DLs.  This
> spot originated in SM, not E5.  The dropped one before that was from the
> call SP0O which had some bad words in Polish directed at an SP station. 
> This spot's IP said DL, not SP.

The subject of filtering is always a vexing one. There is a tradition in 
the US (amongst other places) that nodes provide a "local" view of spots 
whether that be by call area, zone(s) or state(s). Where as here in EU, 
the general tendency is for WW spots. So in this environment, I am going 
to sit firmly on the fence, painful though that can be on the buttocks, 
and suggest that if Lee wishes to inject spots from whatever reasonable 
source, in the obviously thoughtful way that he does, filtering them as 
he thinks reasonable (whether by IP address, content or any other 
metric) - I shan't stop him or discourage him. If a real problem arises, 
in the future, I am sure that Lee will do what's necessary to ameliorate 
it.

But, by the same token, I won't be connecting to the WWW nodes and I 
won't be filtering spots by IP address. Although I could well provide 
the IP address filtering facility if there is a significant demand. I 
will not provide WWW node connectivity, as there only needs to be a 
maximum of two or three nodes providing this and Lee and K1TTT have 
already cornered that market.

> 
> I am not running a Spider cluster, but I will not be implementing
> registration.  It is simply too much work.  In the last 3.5 years I have
> received PC41 data from over 62,000 different calls.  That is how many
> different cluster users there has been.  I estimate a minimum of 10,000
> of them have connected to my cluster.
>

That is, as I think you will be getting the drift of by now, entirely up 
to you. I don't either, as it happens.

Dirk G1TLH