On 2014-05-07 16:55, Per Jessen wrote:
Carlos E. R. wrote:
But the issue, to me, is not really the huge memory, but that the child was not restarted after processing, if memory can not be freed.
Ah okay - I didn't realise that was the issue. That sounds like something to report to the spamassassin people.
Yes, memory is there to be used as needed, as long as it is released when not. The max-conn-per-child switch handles that part, I think. Maybe that causes some performance decrease, spawning more often - but Linux is good at that. I can do some limited testing now, I have to prepare for a little field trip. Reporting takes longer. And if the openSUSE package is different than the upstream one (the obsolete man page indicates that), I can't report to them, but to openSUSE instead. -- Cheers / Saludos, Carlos E. R. (from 13.1 x86_64 "Bottle" at Telcontar)