[Dxspider-support] Cluster hangs

[T3]

Okie-dokie
with or without rant, problem still exists
my end which is possibility something else or something adding to the 
issues,
i do agree about the spot, count being higher at other times then now, 
with other contests
but i don't agree with the PC16 and 17 being filtered out,  or filtering 
out anything from other
clusters, what was the porpoise for them, the PC16-17 that is,  filter 
them out?  why have them at all..
i don't know where the 400 clusters are or the 3K of users.
i have seen that, sense i made that mistake a time back.. (oops)

its starting to look like the issues are with over worked nodes belching 
data over to others
on back-bone networks and making partner nodes choke, possibly bad WiFi 
back-bone's
or partial connected WiFi back-bone's that have been spotted in the past

and the high usage of the dupe file,  from nodes linking to others that 
are already
linked to the network some ware, that needs to be stopped. and enforced.
the multi loop networks that is.

Im ok till i get belted with a excess of traffic, then im trying to 
process, thus  loading
up for another,  Soo the slinky affect, like a bad traffic jam in Boston Ma.

I think some of the problem is growing pains, we add more bell/toys to 
accomadate, mass upgrade, then
have mass issues.. without fully testing under contest conditions. our 
heaviest loads

I don't believe is processor speed of any of that stuff. but i think u 
found a issue in the
debug.files, im looking also but all i see is waves of data, then normal 
operation, then waves of data again
like a slinky or inch worm,

and in some cases im disconnecting the few that i have on the AX25 
ports, and beating up the 9.6k
test cluster i built using yet another windows box,

im testing a windows system so stripped down, with no drivers and any 
excess hardware,
and even that one has issues with the in rush or *waves* of data. so 
this tells me that even the 9.6k
link is over powering it. I have a older copy of Spider, im going to 
change the test box and see
if i can reproduce it.

Sri, just my rant, or help, how ever you see it.


Tony
N1LDY


charlie carroll wrote:

>Eric:
>I'm not convinced it has anything to do with spot rates.  We've had
>rates higher than the current ones when propagation was better.
>
>Caution -- the following is a rant on my part.
>
>Have we reached a point where this hopelessly fouled network needs to be
>re-examined?  Is the issue gettings spot through, or the fact that
>DL1??? logged into the DL9??? node.  (DL callsigns only used as an
>example.)  I personally do not see the sense in propagating 400+ nodes
>and 3000+ users.  Even if I have input filters set up, I still have to
>devote CPU cycles to process the filters.  
>N1LDY just commented on the rate of PC16/17s.  Tony why don't you ask
>the sysop of the nodes that connect to you to stop the PC16/17s?  
>Better yet, why don't sysops be responsible for what they pass on to
>other nodes?  If someone asks to connect a node to you, the other part
>of the conversation needs to deal with the data that's going to be
>exchanged.
>
>Rant over...
>
>73 charlie, k1xx/4
>
>_______________________________________________
>Dxspider-support mailing list
>Dxspider-support at dxcluster.org
>http://www.tobit.co.uk/mailman/listinfo/dxspider-support
>
>  
>