On Monday 01 September 2003 23:59, pineapple wrote:
> This problem has been widely reported in the past, the
> developers HOPE that nextgen routing will fix this.

I am running the latest unstable release that has NGR in it, according to the 
latest report (6163). It is with this release that the problem is being 
experienced.

> Right now your options are 1) put up with it, 2) run
> an experimental version,

I am already running an "experimental" version. The problem did not go away.

> 3) shut your node down or set
> it to transient (if you do go transient, you'll have
> to wipe out your node file to change your node's
> identity because the network now knows about your
> node).

It doesn't bother me that the network knows about my node. What bothers me is 
that no matter what I do, the node gets stuck with threat overloads and 
sometimes routing time overloads far before it reaches the point where the 
allocated bandwidth is consumed. The problem even appears to be slightly 
worse with NGR in 6163 than before with 6162. But, I guess maybe the network 
doesn't _need_ all of my bandwidth at the moment... It just seems odd that 50 
simultaneous transfers cannot seem to deplete 64KB/s.

> The problem seems to be caused too many
> freenet.node.states.data.DataStateInitiator threads.
> A suggestion to the developers; if there are too many
> freenet.node.states.data.DataStateInitiator running,
> just kill some, starting with the threads that have
> been idle the longest.  At least this way the node
> will be able to fullfill some requests instead of
> doing nothing for long periods.

But are those threads doing something? Are they crashed? Or are then just 
waiting for the other nodes to respond? If it is the latter, then isn't that 
a great way to DoS nodes? Open a connection then negotiate it _really_ 
slowly? Maybe there should be a timeout on such things. If a node cannot 
manage to negotiate the relevant operation in, say, 10 seconds, then chop the 
connection anyway, as the negotiating node is just dragging the network down 
with it's own congestion... Sort of like remote node load balancing. ;-)

Gordan
_______________________________________________
Devl mailing list
[EMAIL PROTECTED]
http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to