Marek Kozlowski wrote: > My system is not overloaded and - in many aspects quite typical one.
Great! I believe you. But then there was absolutely no reason to double a limit that would never have been hit. If the system is not overloaded the the initial fork() will always succeed and there will never be any retries. Not five and therefore not ten either. > If I had to double some value something very strange must have > occurred. I'm wondering what was it (a black hole in my memory). That you did double that limit that was not getting hit (not getting hit because your system was not overloaded) then it changed something that had no effective change. In that case it was a noop. Because you were not hitting that limit. Because your system was not overloaded. And therefore there was no effect by changing it. That it was changed was likely simply a misunderstanding at that time. But now that it is later and we all have had more sleep since then and looking at it fresh we can see that it was simply a mistake. No big deal. Nothing to see. I make mistakes all of the time! And now that we know we should remove it from the local configuration and let it drop back to the default as part of the cleanup back to the mainstream. Bob