I've since noticed that I'm still getting numerous connect requests from a large number of remote Freenet nodes. Obviously, I'm rejecting these as the software's not running anymore.
My (mostly rhetorical) point is that it would seem to me that nodes keep information about other nodes for far too long. I'm sure that in this scenario the odd mistaken request would be normal, but it looks a lot more like my node is very much expected to be in service. Is it possible the general reliability of Freenet could be improved by keeping remote node information for less time?
I'm no expert, but I would suggest that Freenet's willingness to keep plugging away at a "down" node should reduce with time. That is, if you haven't had a successful transaction for a while, then leave it alone for a few hours. If you still don't get anything after say a day, then chuck it out of the peer list (permanently). Clearly, if that node contacts you (and is not transient) then put it back in the peer list, pass it around or whatever. Thus, having a peer list of a hundred nodes, 80 of which are down is less likely to cause non-connectivity issues when trying a TTL of 20 or so.
Just my 2 pence - hope it's not too off base(!)
...Ralph Bolton
_______________________________________________________________
Get the FREE email that has everyone talking at http://www.mail2world.com
_______________________________________________ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
