Ian Clarke wrote: > Toad wrote: >> Currently the situation, even with the recently integrated probabilistic >> rejection, is as follows: >> We start off with no load >> We accept some queries >> Eventually we use up our outbound bandwidth, and due to either >> messageSendTimeRequest or the output bandwidth limit, we reject queries >> until our currently transferring requests have been fulfilled. > > Your solutions all seem to be addressing the wrong end of this problem > by accepting that this situation will happen and trying to make NGR deal > with it - but the real question is: Why are nodes getting themselves > into a situation where they have to QR solidly for periods of time? The > QRing should gradually increase with time such that the situation > described above doesn't occur.
Because nodes learn that we're overloaded far slower than we _get_ overloaded. Probabilistic QR should help some, but I think the best thing we've got is Ian's idea (which is not far off from Martin's): Let nodes know what we can handle, and then give them reasons not to try to send us (much) more than that. No, it's not entirely leech-proof, but it's still a good bit better than we've got. What we have here, is a failure to communicate. ;) --hobbs _______________________________________________ Devl mailing list [EMAIL PROTECTED] http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl
