Matthew Dillon wrote:
> Well, you are welcome to run software on whatever platform you feel
> fits it best, but no magic sysctl is going to make your systems more
> dependable if the software running on them is trying to use more
> resources then are available. Unless you are actually running yourself
I want to prevent services that allocate memory only once to a
configurable limit from being killed by the OS because other services
ate it. I want predictable memory failures. Either give me physical
memory, or return an error code. If I see an error code, I immediately
exit. Don't lie to me about physical memory and kill me in the future.
That's what I want. That's all. I may not even _want_ to use swap if the
problem at hand has strict security or performance requirements. In
other words I want some processes to be far more resilient about memory
than others. I can't do it with overcommit.