On Thu, Dec 13, 2001 at 06:53:51PM +0100, Sebastian Sp?th wrote: > We just discussed this on IRC, taking it to the devl list: > > Normally a request times out according to the timeOut setting. But when > Fred receives a "restarted query" message, it will wait even further. > > It might be a nice attack to set up a node and have it answer each > request with an infinite series of restarted query messages, thus > stalling the request forever.
The Freenet kindergarten security corps strike again! :-) > The solution might be to > a) abolish restarted query messages allthogether (thelema) or/and You obviously can't have every node in the chain restart the request at more or less the same time. > b) time out the request "timeOut" seconds anyway, whether there are > pending restarted queries or not... You obviously can't have every node in the chain restart the request at more or less the same time. > Does this make sense, or did I simply work too much today? No it doesn't. I was obviously aware of this when we first implemented the restart. It is not considered a problem because: a) It cannot go on "forever" as the HTL is decremented every time the timer is restarted, and will eventually reach zero causing a Timeout. b) Freenet's structure is handles this form of attack well because Nodes that don't respond correctly eventually loose references. The only way to do this better would be to add limited branching to try to ensure some redundancy - something that is on that long list of things that should be tried in some later version. <> -- Oskar Sandberg oskar at freenetproject.org _______________________________________________ Devl mailing list Devl at freenetproject.org http://lists.freenetproject.org/mailman/listinfo/devl
