For what it's worth I've tried Dave's suggestions here, replicating what he did by dropping connections to 256 and increasing threads to 300. The node comes out of nearly constant overload, instead hovers in the 80 percent range. Also while less total data is being transferred (60MB/hour vs. 100MB/hour) the queue of data waiting for transfer has dropped from the absurd 80MB waiting to squeeze through a 15k pipe over 100 connections range down to a level that varies between 1 and 5MB running through maybe 19 connections.
The CPU load average on the node is now absurdly high. I don't really care about that, but if I was trying to use the box for something else it would be kind of annoying. Maybe some kind of useful hint here? I wouldn't know... _______________________________________________ Devl mailing list [EMAIL PROTECTED] http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl
