Derrick J Brashear <shadow@dementia.org> writes: > a) master should not die if a client hangs up prematurely and it gets > SIGPIPE Right, fixed. > b) slave should not cause me to end up with a machine with a 16k database > when it started with a 17mb one Not really. Is it a problem of getting interrupted in the middle of a transfer when doing full transfer or by some logical/state error? /assar