This strongly suggests that routing is not doing terribly great. The suspicion is that there is some form of network fragmentation going on. Ian refuses to admit the possibility of network fragmentation, so I am wondering how to produce extraordinary evidence to support my extraordinary claim :).
so Ken Corson wrote:
Toad- just count the number of rejections experienced by a single request as it is handled by Node. When you see numbers around 15-20 retries (out of maxRoutingSteps), it will be crystal-clear why things are smelling rather putrid :<( If that turns out to be the case. I'm angling to earn some "developer" points, without yet writing any code :)
then Toad wrote:
So you are suggesting that the reason that routing sucks is that nodes constantly reject 80%+? Perhaps you are right, but from what I have seen the rejection is mostly because of bandwidth usage... is that the consensus? Why do your nodes reject, people?
Yes, this is what I am suggesting : If 90% of requests are rejected, everywhere... then the requesting node has to retry, using the "next-best" node, and then the "next-to- next-best" node, then the "notta-so-goodd" node, until finally, our request winds up accepted by one of the LEAST APPROPRIATE nodes.
Remember how you "recently decided" to stop querying nodes in the wrong order, from worst to best candidate ? You know, that major bug you found and fixed around 6275 or so ? I was thoroughly impressed by, and grateful for, your discovery! ... which I jokingly termed the "Toad took his foot out of everyone else's mouth" event. Well, "IIITT'S BAAACKK... ," only in a better disguised form.
The only reason any content is found and moves around at all, is because some small percent of requests actually arrive where they wanted to go in the first place. And users are kind enough to retry their requests 5 to 10 times to see if it hits something ...
This is my theory, derived from watching everyone else and looking at my node's statistics. I hope this is useful and pans out. This just describes the problem. It doesn't offer a solution. But if it turns out true, i still want some points :) I'm just gonna keep bangin' on this QR-crap until someone convinces me to shut up.
Ken
_______________________________________________ Devl mailing list [EMAIL PROTECTED] http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl
