Hello Theo, On Fri, 17.03.2006 at 16:44:58 -0700, Theo de Raadt <[EMAIL PROTECTED]> wrote: > Then the entire idea is to improve the drivers to erradicate this > special treatment that makes them believe that they should syslog > of kernel printf us to death. That is not the solution you want.
this is a bit confusing to me. Is "improving the drivers" or "printf us to death" not the solution I want? Either way, if anything is really logging someone to death, then there are probably bigger problems than copious logging ahead. Ie, normally, such a state transition on a WAN link should occur at a frequency of less than once a week, but rather once or twice a year. I can't imagine how such infrequent events could lead to system or administrator overload due to the volume of logged data. But then I still may be missing something. > Fixing this will require a framework. Not just kernel printf. Do you have something specific on your mind? Best, --Toni++
