Please read this one. Think he makes sense. If so we can remove the global estimator from the eq, with Martin's formula we should not need it. This could also explain why
we are seeing such high numbers of requests on nodes - routing thinks they will react
fastest and they do by rejecting...
Not unless the algorithm has been mis-implemented.
Routing something to a node that is almost certain to fail shouldn't happen in NGR, since the otherwise attractive short time required to fail (tFailure) will be dwarfed by the very unattractive large amount of time required to get the data from another node in the estimate.
Remember, NGR's estimate is always "how long will it take me to get the data or receive a legitimate DNF". It never estimates "how long will it take me to get a failure". Therefore a failure will always contribute the time required to retrieve the failure PLUS the time required to successfully re-request the data elsewhere to the estimate.
If we are forgetting to add the re-request time then that would be a very serious bug. I should probably RTFS to make sure that this isn't the case.
Ian.
_______________________________________________ Devl mailing list [EMAIL PROTECTED] http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl
