[Dxspider-support] Inconsistent spots - long message

Dirk Koopman djk at tobit.co.uk
Tue Sep 11 01:25:09 CEST 2007


charlie carroll wrote:

> The root of all evil turned out to be spots from WR3D!  But why?  It 
> finally dawned on me that this was the isolated connection.  So, I 
> returned to a normal connection and voila, spots from WR3D no longer go 
> missing.

and now I have reread this in more detail, we see artifacts from the 
"good old days" reappearing!

> 
> The missing spot problem is easily re-producible each time I isolate the 
> link to WR3D.  At this point, I'll toss this little gem into your lap 
> and await any questions.
> 

Set/isolate is designed to be draconian. "Sh/filter node_default" is the 
thing to look at together with "sh/filter wr3d". You must have "acc/spot 
wr3d all" and "acc/spot node_default all". Because, otherwise, even 
though spots appear to *your* users from wr3d, they are not transmitted 
onwards and (possibly worse from your point of view) will dupe block 
those same spots coming in from elsewhere.

With set/isolate it does just that. See:

http://www.dxcluster.org/main/adminmanual_en-1.html#ss1.12

for some words. But the upshot is: unless there is egress filtering in 
place (as above), then stuff will not cross the boundary set up by the 
isolation EVEN THOUGH you and your users will see everything that comes 
in from both directions. This is there for your benefit, not confusion 
(honest).

Getting the filtering right should cure the problem.

Dirk



More information about the Dxspider-support mailing list