Hi folks,

We're working to implement a remote queuing feature which will hopefully increase the number of successful downloads over time.

Avi

an interesting experiment to try is to use shareaza and LW immediately one after another (or concurrently I guess if you have the grunt) - what I (at least) find is two interesting things - first shareaza connects to the same sorts of clients except that BS connections are common, stable and lead to downloads (unlike LW) and two, the ultrapeer connections remain fixed for FAR longer than is the case with LW - not entirely sure which setup leads to the best results 'at the end of the day' and I prefer the LW layout/design but it does suggest that LW might usefully give attention to the ultrapeer connections cos the greater ultrapeer volatility with LW can't be just attributed to 'thats just how the network is'

Cheers, bad_vlad

The higher ultrapeer volatility is a result of the ConnectionWatchdog function LimeWire has. If you don't receive any messages from a connection for twenty seconds or so, it is kicked. LimeWire does so, to identify bad hosts quickly. It has the interesting side effect that your ultrapeer connections seem to be kicked a lot more often if you are sharing no files.

Shareaza also monitors connections and removes them if they are not performing, but it takes into account the traffic it would expect to get from its neighbours and tries not to remove a connection unless there is actually something wrong with it (as opposed to a traffic lull).

OK - the connection Watchdog function is sensible and smart - but - riddle me this - in LW I have more than 800 files available for sharing and it'd be a rare moment that the available upload slots weren't fully occupied which would seem to reduce ultrapeer volatility - conversly - in shareaza I chose not to be an ultrapeer (56k dial up connection), lots of my shared files were greyed out for some reason, and because I'd only just connected I was designated a 'doubtful' host because there was no record of any uploads which would seem to increase ultrapeer volatility - and yet - I still got far more stable ultrapeer connections that included BS and (crucially) they 'worked' -- is it that the 20 secs in LW is problematic or that the shareaza monitoring that allows for 'lulls' is better in practice - whatever - the end result is that shareaza threatens to threaten the dominance of LW I reckon

Cheers, bad_vlad

The algorithm LimeWire uses is very strict and at first 80% of the connections may not last longer than a few minutes, especially if you are keeping more than 3 ultrapeers connections. - However this ultrapeer volatility is not necessarily a big issue. The longer you are running LimeWire the more stable the connections become and after a while you can be pretty sure to be connected to 'good' ultrapeers.

OK - my resistence is dissipating - but just to completely put me out of my misery - why can I succesfully download via a BS ultrapeer in shareaza but never in LW - and - still not entirely convinced about 'good' ultrapeers eventually settling in for a respectable time - instead what I find is that the volatility actually increases and the only solution is to exit LW and open up again to take advantage of the burst of connections I start with that then diminish to the point of complete disappearance

still cheery: bad_vlad

All ultrapeers do is relaying messages, like queries or search results. They have absolutely nothing to do with your downloads.
If you keep restarting LimeWire you will not get many stable connections. The best approach is, to leave it running for a while. If LimeWire doesn't seem to stay connected on its own (without reconnecting it), your uploads are probably using up most of your outgoing bandwidth, so most of the incoming gnutella messages are dropped (that can cause the connection to be dropped after a while) or you are keeping too many gnutella connections which might cause the same problem.

I do realise that the ultrapeers are just links - my point is that BS links work for me in shareaza but never in LW and I'm curious why that should be (I think its something to do with code whereas you seem to think its spomething to do with the way I use LW

bad-vlad

I don't understand your problem. LimeWire does connect to the new BearShare 4.0 ultrapeers without any problems. The old BearShare nodes aren't ultrapeers, so LimeWire will not connect to them while in leaf mode.