On Thursday 31 October 2002 10:34 am, you wrote:
Questions regarding the correct method to stop and restart a node. This
morning I upgraded to 604 and started it, and it runs very well.
I stopped the node to reload some relaxed paramenters in freenet.conf and
then restarted it, all within about 30 seconds or a minute. As soon as the
node became active, it spawned large numbers of java processes, sucked up
all CPU time and the pipe here became jammed.
I stopped the node, then took a look via tcpdump at what was coming in.
Lots of orphaned traffic from other nodes.
Is it possible that if a node is restarted too quickly it will become
confused by other nodes attempting to continue sessions that were started
in the previous instance of the node? If that's the case, is it reasonable
to assume that the minimum announce delay for a persistent node should be
longer than the maximum amount of time other nodes will attempt to continue
sesssions with a prior instance of the local node?
Replying to myself with results of a little experiment. I left the node down
for about an hour, then restarted. Even before the announce interval had
elapsed the node was once again swamped with connections, as verified by
tcpdump since the proxy interface was not yet available. Again, this is 604
which was running great before I stopped it briefly.
OTH while the CPU is being hammered I can still use the lan here to get out
on the net. So in general things are still better with 604.
___
support mailing list
[EMAIL PROTECTED]
http://hawk.freenetproject.org/cgi-bin/mailman/listinfo/support