[Dxspider-support] PC41 loop?

Dirk Koopman G1TLH gb7tlh at dxcluster.org
Thu Jun 17 14:17:38 BST 2010


On 16/06/10 19:30, Lee Sawkins wrote:
> Hi Dirk
>
> What you have done with PC9x messages works great. If all the other PC
> messages could be redone to have time tags, this would go a long ways
> towards cleaning up the backbone traffic.

I agree, but I don't particularly want to replace the whole lot, I just 
don't have that amount of spare time to make sure that I can remain 
upward compatible. It is probably the "upward compatibility" that is the 
most time consuming part of fiddling with the protocol.

>
> At times I see alternating PC41s with name information. One PC41 for a
> call will have the name as "Bob" and it will immediately be followed by
> another one with the name "Robert" for the same call. Accented letters
> in names or QTHs get changed to "?". We then have alternating PC41s with
> different content. This just keeps looping.

That will die down as people either increase the PC41 dupe time or upgrade.

>
> I set my location as 49 11 N 122 30 W. This goes out in a PC41. I see it
> come back alternating as 49 11 N 122 27 W and 49 11 N 122 30 W. How does
> this happen? This resets the dupe filtering as well.

This is a problem with different authors using another reference point 
and then enforcing their point of view when they receive a locator. If 
they simply left the lat/long they have alone it would not be an issue.

>
> I also see alternating PC19 and PC21 from the same node. This goes on
> for hours and hours, one right after another.
>

Why are you seeing PC19/21 at all? I have not seen any for years (except 
from the odd AR-C or AK1A node).

One of the important changes I have made with PC9x is that I ignore 
routing stuff from directly connected legacy nodes that originated or 
refers to other nodes that might be connected to that legacy node. 
Essentially I only accept PC19 and PC16/17 (viz: the node info and only 
its users) for that node alone. The rest I just ignore.

This means that there is effectively only a single level of legacy node 
connected outside the PC9x speaking cloud. As far as I am concerned, 
what else (in the way of routing info) they send to me and amongst 
themselves is so much hot air.

> I also at times see DX Spots come in with a hops count of H1. This stops
> the spot from going further. Sometimes I then immediately see the same
> spot arrive with a hops of say H93. I process this "dupe" spot and send
> it on as H92. Does Spider do this? I don't think so. If not, it means
> that a cluster that sets all its DX Spots to a hops of H1 can stop other
> clusters from passing the spot further. I have one cluster that sends me
> almost all spots with H1. I do not start dropping DX spots until I see
> an H10 or higiher.
>

This is an old complaint of yours that we are both aware of. However, in 
the modern circularly connected world it is irrelevant. The fact is 
that, if you receive it, your users will get it. Most (maybe all) PC9x 
nodes have two or more partners. As you say above, if you get a spot 
with H1, you will likely get the same spot almost immediately from 
somewhere else with a higher hop count.

But means that so will other nodes. So if you can't pass on your H1 spot 
to your partners, because they are connected via other nodes, they will 
get the H93 (or whatever) version from somewhere else. So they still get 
it, just not from you.

Oh, and having looked at that bit of the code for the first time in 
years, I see that I no longer automatically re-hop sentences of any kind 
without some kind of re-hopping filter in place. The hop_table is 
otherwise only used for sentences that a node originates.

Dirk

PS I am now sending PC61s to DXSpider (or compatible) nodes for 
originated nodes.





More information about the Dxspider-support mailing list