Oh, there might be a correlation between the ammount of connection gnutella has to the Gnutella-net and the time it takes till gnutella comes down.

For me it took at least some hours with 6 connections, but can happen as fast as five minutes with 10 conections.

Do any of you see the same?

cu Martin

I'm having the same problem with the OSX and PC versions of Limewire so I believe it's OS independent. Maybe it's a Java thing.

If this is OS Independant we should definitely get in touch with the developers so that _really_ anoying Bug can be fixed.

I´l try to confirm that thing on a pc too.

One question left: Which Java VM do you use? 1.4 or 1.3.x? That might be verry important.

Thanks a lot,

cu Martin

I am still getting this crash average once per 2 hours
 

at least.

I'd bet it's in the gui -- some kind of event queue overflow or deadlock -- because I can hammer networking on my machine (run iSwipe & eDonkey & several remote shells and launch many http connections) and it doesn't seem to affect the frequency of the crash.

I'm going to try fern.

I have a g4 450dp -- I'm curious to know if this crash is happening more often on dual processor machines.

A bug, that could create hundreds of threads after running LimeWire for a certain time increasing memory and cpu usage drastically, has been fixed in the 2.4.0 code.

Ah, I see. I have let run the latest version from CVS (no version in CVS, so I named it 2.4.0jum1) for all night and it did not lock up. The threads fix appears to have done it.

This is great news!

Sorry to ask this here, but could you send me that working build per e-mail?

All my attempts to build this thing from cvs ended in overflowing error messages. :(

mhaecker at mac.com

Thank you a lot!

cu Martin

The 2.4.0 beta has just been released precompiled on the LimeWire homepage.

http://www.gnutellaforums.com/showth...threadid=10970

I have also done a build that is available at:

http://baghira.han.de/~jum/LimeWire%...jum2%20OSX.dmg

Thank you guys a lot!

Your help to make me go Gnutella again is _much_ apreciated!

:)

Thanks, Martin