I have only acted as an ultrapeer since upgrading to 2.9.8 six days ago.

Fact: every time this has happened, the entire system has crashed (and in 9/13 crashes, it has even required rebooting the DSL connection to reinstate the default settings for the internet connection).

Fact: the longer my machine runs as an ultrapeer, the more slowly it performs until it finally crashes. It literally grinds to a halt.

I have performed the ritual clensing of the stables and reinstalling LW, but this has made no difference. My inclination is to be supportive of the network. It seems to me to be a part of the price to be paid for access to the whole. But unless I can solve this problem of performance degradation, I shall be forced to disable the ultrapeer function. I suspect that it must be a problem of churn overload — that the memory allocation to the message routing and traffic management portions of the package tries to expand to dynamically record the sometimes violently fluctuating traffic but slowly overloads??? You tell me. Whatever the problem is, it is actively preventing me from leaving LW on in the background as I try to get on with other work.

All advice appreciated.

In my last session as an ultrapeer, LW crashed four times, taking the OS with it twice, so since your attitude seems to be either don't know or don't care, that's going to be my attitude from now on. I am disabling the ultrapeer function. Being a leaf is best.

But consider. If every public-spirited unlimited 512 account holder like me withholds his spare bandwidth capacity from the rest of all you takers, how long will your freeloading continue?

Are you both running on OS X? We've made some improvements in 2.9.9 that should improve the situations, but the 1.3.1 Java implementation has several fatal bugs that are outside of power to solve. We're hoping that the 1.4.1 Java release will solve these issues, and LimeWire will be using that shortly.

Any more information you can provide regarding these crashes would be helpful.