[Dxspider-support] Forgeries

Wed Mar 12 19:12:17 GMT 2025

 This is helpful. I still don't understand why H is reported if it is arbitrary. What does it affect? What is it used for? What breaks if it is missing?
So let's assume that extra data is not a problem as your list of nodes for each hope will be just 5. Can you explain the actual algorithm that you will apply to determine if the spot is good or bad? I don't understand how that data will be used in practice. Just use a plain English or Italian pseudo code to explain your idea:   
   - For each spot, I will create a list of all nodes that have routed the spot.
   - Then I will examine each node and do this and that
   - If this happens, I will do A, if that happens I will do B.
   - Spots will be dropped based on such and such criteria.
   - I am making such and such assumptions
Rudy N2WQ
    On Wednesday, March 12, 2025 at 02:55:58 PM EDT, IZ2LSC <iz2lsc.andrea at gmail.com> wrote:  

 For PC61 originated by dxspider the default hop count is 30.I cannot comment for spots originated by others.I can see on my node PC61 originated by VE7CC coming in with H95
Anyway, Kin EA3CV did an excellent paper analyzing the network and he found that each node can reach any other node in 2-5 hops on most of the cases.So I do not expect to see long path recorded as you depicted (as far as there are no loops, but this is a good way to discover loops indeed).
73s
Andrea

-->

On Wed, Mar 12, 2025 at 6:44 PM Rudy Bakalov <r_bakalov at yahoo.com> wrote:

 350 bytes per hop per message, not 350 bytes per message. It's give or take another 25 MB for 80,000 messages.
If H is not what PC61 says it is, what is it? If it is decremented, what is the starting value to get to 98? And what is the purpose of H if unreliable, why send it at all?
Rudy N2WQ
    On Wednesday, March 12, 2025 at 01:34:47 PM EDT, Kin <ea3cv at cronux.net> wrote:  

Average spot value (60/130) 95 bytes. But I will use the worst case, 150 bytes.

Received frame size:

14(Ethernet header)+20(IP header)+20(TCP header)+150(Payload)+4(CRC) = 208 bytes.

80000 messages/min. approx. 1334 messages/s

1334 messages/s * 208 bytes * 8 bits = 2219776 bytes/s

Necessary line speed = 2.22 Mbps

With the assumed increase of 350 bytes per message:

14(Ethernet header)+20(IP header)+20(TCP header)+500(Payload)+4(CRC) = 558 bytes.

80000 messages/min approx. 1334 messages/s

1334 messages/s * 558 bytes * 8 bits = 5954976 bytes/s

Necessary line speed = 5.96 Mbps

But if instead of having 35 neighbouring nodes, we had 10:

(5954976 bytes/s * 10) / 35  = 1.70 Mbps  

Conclusion, pecata minuta for the connection to the provider and a waste of CPU/Memory with so many links.

Ah, we take the value of about 5 hops as valid when we are interested, now we should not use 50.

Kin EA3CV

PS

20 CCC + 15 DXS

De: Dxspider-support <dxspider-support-bounces at tobit.co.uk> En nombre de Rudy Bakalov via Dxspider-support
Enviado el: miércoles, 12 de marzo de 2025 16:57
Para: iz2lsc.andrea at gmail.com; The DXSpider Support list <dxspider-support at dxcluster.org>
CC: Rudy Bakalov <r_bakalov at yahoo.com>
Asunto: Re: [Dxspider-support] Forgeries

Let’s think this thru:

1/Logs show spots with H as high as 98! Recording the node of each hop will increase PC61 payload on average 6 characters per hop. Let’s say average H is 50, that would be 300 characters plus the delimiters or about 350 characters.

2/My cluster received over 80K spots per minute. Imagine the traffic created by 80K messages with additional 350 characters per PC61 message.

Can you describe the full algorithm you are proposing? How are you going to use the additional data to make a decision to accept or drop a spot?

When comparing PC61 to BGP remember that BGP is designed to support two-way data flows. PC61 is strictly one-way and doesn’t need routing overhead.

Rudy N2WQ

Sent using a tiny keyboard.  Please excuse brevity, typos, or inappropriate autocorrect.

On Mar 12, 2025, at 11:19 AM, IZ2LSC via Dxspider-support <dxspider-support at tobit.co.uk> wrote:

Mike, this is why I'm saying that we need to record the whole path, not only the last hop.

Every node has to append it's ID.

Andrea

-->

On Wed, Mar 12, 2025 at 2:05 PM Mike McCarthy, W1NR via Dxspider-support <dxspider-support at tobit.co.uk> wrote:

The current release does tag spots from the incoming node partner with a 
"via" tag. You would need to analyze all the logs in between, using the 
"via" tag to trace it back to the source.

On 3/12/2025 8:06 AM, IZ2LSC via Dxspider-support wrote:
> We are all senior developer with AI!
> At least you agreed that what you proposed for partner identity is not a 
> solution against the flooding.
> Perhaps do you also agree about the need to record the spot path?
> 
> 73
> 
> Andrea

-- 
73 de Mike, W1NR

THAT was the equation. EXISTENCE!... SURVIVAL... must cancel out... 
programming!

- Ruk -

_______________________________________________
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/20250312/71dd2424/attachment-0003.htm>