If an operation takes longer than the client idle timeout, then nntpcache will erroneously idle out the connection. I don't have the source to hand, but adding an "alarm(0)" after reading the command from the user will reset the timeout. As an aside, I've got the patches from Aaron Nabil against 1.0.7.1 which reduce the instances of "Memory Corrupt" problems, but they still occur. Is the only way to track these down to turn on news.debug logging, and try to correlate errors with previous operations? The other problem I am seeing in my logs is "unlinking short xover file" errors. This is under Linux 2.0.30, and the other patch to nntpcache is my modification to use select to detect dead server connections. -- `O O' | Nick.Holloway@alfie.demon.co.uk http://www.alfie.demon.co.uk/ // ^ \\ | Nick.Holloway@parallax.co.uk http://www.parallax.co.uk/~alfie/
