[Dxspider-support] Re: Inconsistent user reporting on different DXCluster nodes

Fri Jun 18 22:35:02 BST 2004

No, sorry.

Or at least not directly. The actual problem is to do with a load of
"rspf" checks I put in and also the fact that routes which seem good
ones at node A may prove to be dead ends when the talk gets to node D.
In a looped network, talks (and similar things) have to, at least, be
output on every interface that seemingly provides a route (or just
broadcast, which is effectively what you end up with, anyway, in a
looped network).

Trouble is: this will create even more havoc with other "legacy" (and
even quite modern) cluster software using PC protocol - 'cos it ain't
designed to cope with that. 

I did actually think about using the chat logic, recognising callsigns
as meaning 'to me' as a means of achieving this. It may now be safe to
do that as chat seems to be used quite a bit and <tin hat><body armour>
I havn't had a serious complaint about it for at least 6 months<body
armour></tin hat>.

Dirk

On Fri, 2004-06-18 at 19:38, Kjell Jarl wrote:
> Hi Dirk,
> While you are at it, is the talk problem I mentioned the other day ralated?
> 73
> Kjell
> 
> 
> Dirk Koopman wrote:
> 
> > After this conversation with Martin G0HDB, I had another look at the
> > PC17 handling code. I have made some changes such that, with a following
> > wind, a few more PC17s will be sent on than previously. 
> > 
> > I would appreciate it if those of you on CVS would update and restart
> > and then monitor the situation to see if the user map starts to bear a
> > better resemblance to reality than has done in the recent past.
> > 
> > Dirk G1TLH
> >  
> > On Thu, 2004-06-17 at 19:19, Martin Davies G0HDB wrote:
> > 
> >>On 17 Jun 2004 at 17:28, Dirk Koopman wrote:
> >>
> >>
> >>>On Thu, 2004-06-17 at 17:06, Martin Davies G0HDB wrote:
> >>>
> >>>
> >>>>As previously mentioned to you there would seem to be something
> >>>>amiss with the way that user (and node) add and delete PC messages
> >>>>(PCs 16, 17, 19 & 21) are being handled, resulting in users and
> >>>>sometimes nodes remaining shown as connected even when they have
> >>>>been disconnected either by positive action or through timeouts. 
> >>>>Any thoughts/comments on this interpretation of the evidence?
> >>>
> >>>Martin
> >>>
> >>>I would agree that there is something wrong. I suspect it is
> >>>specifically PC17 handling that is the issue. I have stared long and
> >>>hard at both the code and at passing traffic and I cannot see what is
> >>>going wrong. There appear to be circumstances when a PC17 should be
> >>>passed on - and isn't.
> >>>
> >>>Whenever I do tests it seems to work OK. I suspect this is a loop
> >>>withing loop problem made worse by some of the route filtering that is
> >>>going on. 
> >>>
> >>>Should you come across a set of circumstances that you can say might
> >>>be a cause, I will gladly look at that.
> >>
> >>Hi Dirk, many thanks for the reply.  I'll see if we can come up with a scenario 
> >>that gives rise to the 'problem' but I suspect that, as you say, there are so 
> >>many variables involved that it might be difficult to pin down a precise cause.  
> >>However, hopefully one day we'll have one of those 'Eureka' moments and all 
> >>will become clear!!!
> >>
> >>--
> >>73, Martin G0HDB
> >>
> > 
> > 
> > 
> > _______________________________________________
> > Dxspider-support mailing list
> > Dxspider-support at dxcluster.org
> > http://www.tobit.co.uk/mailman/listinfo/dxspider-support
> > 
> > 
> 
> _______________________________________________
> Dxspider-support mailing list
> Dxspider-support at dxcluster.org
> http://www.tobit.co.uk/mailman/listinfo/dxspider-support
-- 
Please Note: Some Quantum Physics Theories Suggest That When the
Consumer Is Not Directly Observing This Message, It May Cease to
Exist or Will Exist Only in a Vague and Undetermined State.