On November 21, 2003 11:23 am, Ed Tomlinson wrote: > Hi, > > When I use a bandwidth forcast to limit request (via QR...) > I noticed the same effect. In my case sendData was about > 0.90 so most of my sends worked. About 80% of my bandwidth > was for trailers and fred was using all the bandwidth I allocated > to it. Looking that the requestSuccessRatio I saw numbers > like .25 and above (real nice...). Looking at routeSuccessRatio > number we about 0.0.15 or 0.02 - which is normal. > > What was happening was my node had most of the data locally > and was happily sending it. I routed very few requests. Part of > the reason for this is that we do not QR if we have the key locally > or if we can DNF from the failure table. > > Suspect a similar effect is happening on your node. On my > current node, with a reset datastore 3 days old, I see: > > Transmitted/Recieved 610M / 710M That should be 710M not 71M
Ed > > On November 21, 2003 09:39 am, Niklas Bergh wrote: > > > At many times I have thought things like 'Where does all this data my > > node sends go to!' > > > > An example.. This is from lostlogics ubernode: > > Total amount of data transmitted/received 16 GiB/3,713 MiB > > > > And I am seeing similar effects on both of my two nodes. With a large > > store the amount of data transmitted is significantly larger than the > > data received. With a smaller store they are closer together. Data > > transmitted has ALWAYS been higher than data received though.. . I > > thought this was more or less the norm. > > > > Today I had a chat with ejhuff on IRC and he showed me his numbers: > > Total amount of data transmitted/received 164 MiB/772 MiB > > > > As you can see his node has consumed about five times more data than it > > has sent to other nodes.. I though that this was because of many local > > requests but he assures me that he hasn't done much local requests at > > all.. > > > > The only other reason I can see for this situation is that his node has > > completed downloads of externally requested keys which he for some > > reason or another has been unable to pass on to the requestor. If > > routing where working properly then these downloaded keys ought be > > requested again by other nodes.. As it is now with routing not working > > properly the chance that his node will be queried for the same key again > > is very low... Meaning that those downloads where nothing but a waste of > > bandwith for him and a waste of upload bandwidth for one or more other > > nodes. Those keys will just hang around his store until they get dropped > > from it because they was never requested again. > > > > Is there anything that can be done to counter this? Maybe cancelling a > > download when the connection we are relaying the data to dies? Or should > > we cap bandwidth to Math.max(outboundBw,inboundBw) or something? I > > assume that if we don't do the former at least then any node can be used > > to perform a DOS against other nodes, No? > > > > > > Do many people see this effect? > > /N > > > > _______________________________________________ > > Devl mailing list > > [EMAIL PROTECTED] > > http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl > > _______________________________________________ > Devl mailing list > [EMAIL PROTECTED] > http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl > _______________________________________________ Devl mailing list [EMAIL PROTECTED] http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl
