[Dxspider-support] Number of hops on a node

Fri Jan 28 15:51:07 GMT 2022

I understand what you are saying Lee, but maybe it is time to bring the
protocol into line with the present and with a view to the future. It has
been a long time since the first AK1A format packet was sent. You have had
to implement new PCs to avoid the inconsistency problems with which AK1A
was created.
The current works, but improving it from scratch, not inheriting problems,
can make the current network nodes more effective and efficient. A new
exclusive interface can be created for the interconnection of nodes and web
applications, logger, ...

You may think that third-party software is one of the stumbling blocks to
dealing with change, but perhaps it is time to stop using those old
developments that are no longer maintained and replace them with others
that are appropriate for the present. As much as I like a program that only
runs on MSDOS or Win95, I don't think I should sacrifice too much of my
machine by maintaining obsolescence. There are plenty of developers in the
community who would surely create new tools as useful or more useful than
the old ones.
I don't see the teams at N1MM, DXLab, Logger32, ... (to mention a few) not
updating their software and being cut off from the world wide web.

As you said, you and Dirk carry the weight of the DXCluster development
nowadays. You both have a lot of experience and I'm sure you could open a
dialogue forum in order to look for a new protocol (say i.e. v2) that could
coexist with the current one but giving it an expiry date, so that the old
programs could be migrated to the new developments.

I know it's a big job, but you guys have shown that you are up to it, and I
don't think you wouldn't like a challenge like this ;-)

We'll keep talking...

Kin
EA3CV

El jue, 27 ene 2022 a las 22:01, Lee Sawkins via Dxspider-support (<
dxspider-support at tobit.co.uk>) escribió:

> Hi Kin
>
> The first cluster server program was Packet Cluster by AK1A.  He quit
> developing Packet Cluster many years ago.  Since then many others have
> written programs that are mostly compatible with the AK1A protocol.  The
> problem is that there is no governing body that defines exactly what the
> AK1A protocol is.  You will find no mention of the Internet or Skimmer
> spots in the AK1A protocol.
>
> Dirk and I are the only ones currently developing cluster server code.  We
> work together to add functionality but at the same time realize that there
> is a lot of legacy software that is in use and cannot or will not be fixed.
>
> Lee VE7CC
>
>
> ----- Original Message -----
> From: Joaquin . via Dxspider-support <dxspider-support at tobit.co.uk>
> To: The DXSpider Support list <dxspider-support at tobit.co.uk>
> Cc: Joaquin . <joaquin at cronux.net>
> Sent: Thu, 27 Jan 2022 12:01:07 -0700 (MST)
> Subject: Re: [Dxspider-support] Number of hops on a node
>
> Hi Lee,
>
> Interesting explanation.
>
> From what I've been seeing, I think there are nodes that have decided to
> lower the number of initial hops, because there is no other explanation
> after analysing their location in the global network with respect to my
> node. I think this is a mistake that makes the concept of hops in a routing
> protocol partly unusable.
>
> You developers have not thought about evolving the structure of some of
> these messages. I am referring for example to those without a source node
> such as PC41, PC20, ... which is incomprehensible in a layer 7 protocol. Or
> as you said, those that do not have a last field with hops.
>
> I am very grateful for the information you have given me, it is difficult
> to find information (the specification) of the protocol and its fields.
>
> Regards
>
> Kin
> EA3CV
>
>
>
> El jue, 27 ene 2022 a las 19:32, Lee Sawkins via Dxspider-support (<
> dxspider-support at tobit.co.uk>) escribió:
>
> > Hops were used by the AK1A software.  They were used for 2 purposes
> > AFAIK.  These were as a Time To Live (TTL) and as a way of limiting
> > distribution of spots.  Some PC messages do not include hops.
> >
> > When I wrote CC Cluster node software I found a couple of serious
> problems
> > with hops or lack of hops.  PC51 messages do not include hops.
> Sometimes a
> > PC51 would be forwarded to all connected nodes only to be echoed back
> from
> > the node that I had just sent the message to.  This caused an infinite
> loop
> > and locked up my software.  I found the only way to stop this problem was
> > to ignore all PC51s that did not include both my node's call and the node
> > call that I was connected with.  I now only use PC51s as a keep alive
> timer
> > and a verification that connected node call has my call correct.  CC
> > Clusters will not accept data from a new node connection until they see a
> > correct PC51.  This will cause the connection to eventually time out
> unless
> > both sides have the correct call signs.
> >
> > I found that DX Spider nodes sometimes increased the hop counts.  During
> > tests I could send a PC message to a Spider node with a hops count of H1
> > and have it come back from another Spider node at a hops count of H24.
> >  Once again this was unacceptable as it meant that PC41 data for instance
> > could and would circulate forever.  They would never time out.  I solved
> > this problem with CC Cluster software by dropping forwarding of PC41
> > messages.
> >
> > When a node receives a PC11/PC61 at  a hops count of H1 it is supposed to
> > send the spot to its users and not forward the spot to any other
> connected
> > nodes.  This means that the other connected nodes may not receive the
> spot
> > at all.  CC Clusters use a different method of determining if the spot
> > should be forwarded or not.  The CC Clusters keep track of which spots
> have
> > been forward and which ones are "hopped out".  If the CC Cluster
> receives a
> > spot that was previously hopped out but this time it has a hop count
> > higher, it will forward that spot to its other partner nodes.  This
> results
> > in fewer good spots being lost because of incorrect setting of hops by
> > other nodes.
> >
> > Dirk constructed PC9x messages with a much better way of determining
> > whether to forward them or not.  These include a "time since midnight"
> > (TSM) number which is unique.  The hops number is ignored even though it
> is
> > there.   Any PC9x message is dropped unless it has an increased TSM or
> has
> > wrapped around at 0000Z.  CC Clusters use this better method as well.
> >
> > Since implementation by CC Clusters of the above, there are no longer any
> > data loops that CC Clusters participate in.
> >
> > Lee VE7CC
> >
> >
> > ----- Original Message -----
> > From: Dirk Koopman via Dxspider-support <dxspider-support at tobit.co.uk>
> > To: Joaquin . via Dxspider-support <dxspider-support at tobit.co.uk>
> > Cc: Dirk Koopman <djk at tobit.co.uk>
> > Sent: Thu, 27 Jan 2022 08:11:29 -0700 (MST)
> > Subject: Re: [Dxspider-support] Number of hops on a node
> >
> > AFAIK "no-one" changes the default values of the hops field. Apart from
> > (possibly) me. And, again AFAIK, I don't change anything.
> >
> > Have a look at DXProtVars.pm
> >
> > Which means that I not using these values for PC92 etc. I don't know
> > why. Neither can I remember why I have "special" hop counts for the
> > sentences in %hopcount - other than to test the mechanism. Which given
> > that I created/touched this file as recently as June 2007 is clearly
> > unforgivable.
> >
> > I have no intention of changing this file. So these are the defaults.
> > Given that I had to spend about 10 minutes looking for this information,
> > I'm guessing that no-one else has either. So for DXSpider nodes the
> > hopcounts for those sentences are correct. For PC92/93 they are all
> > hardcoded at H99.
> >
> > You can probably write any code you might need using the defaults in
> > DXProtVars.pm BUT this will only work for DXSpider nodes. (Nearly) Every
> > other software will likely be using H99. You can probably assume that
> > anything above (say) 70 started at H99.
> >
> > Hope this helps
> >
> > Dirk
> >
> >
> > On 27/01/2022 10:18, Joaquin . via Dxspider-support wrote:
> > > Hi all,
> > >
> > > Is there any way to know the number of hops a node has defined? Does
> > > it show up on any PC?
> > > I have tried sh/hops but I see that it is only useful if you have
> > > redefined the hops locally.
> > >
> > > Regards,
> > >
> > > Kin
> > > EA3CV-2
> > > EA4URE-5
> > >
> > >
> > > _______________________________________________
> > > Dxspider-support mailing list
> > > Dxspider-support at tobit.co.uk
> > > https://mailman.tobit.co.uk/mailman/listinfo/dxspider-support
> >
> >
> > _______________________________________________
> > Dxspider-support mailing list
> > Dxspider-support at tobit.co.uk
> > https://mailman.tobit.co.uk/mailman/listinfo/dxspider-support
> >
> >
> > _______________________________________________
> > Dxspider-support mailing list
> > Dxspider-support at tobit.co.uk
> > https://mailman.tobit.co.uk/mailman/listinfo/dxspider-support
> >
>
>
> _______________________________________________
> Dxspider-support mailing list
> Dxspider-support at tobit.co.uk
> https://mailman.tobit.co.uk/mailman/listinfo/dxspider-support
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.tobit.co.uk/pipermail/dxspider-support/attachments/20220128/9defeb59/attachment-0001.htm>