Strange fork() behaviour

[Martin Preishuber]

Harald Hanche-Olsen wrote:

> | There are some strange things with fork() and python 1.5.2 going
> | on here.
> |
> | I do have some main program which is basically
> |
> | while 1:
> |     fork()
> |
> | The program forks under certain conditions and works as expected
> | but somewhen the main process dies even though it is in the
> | while 1: loop ... there's no error message ... did anyone
> | see anything similar ? the program worked fine with 1.5.1
> | (this is all on a linux box running rh 5.2 and kernel 2.2.6-ac1)
 
> This is a joke, right?  What I would expect the above program to do is
> to fork processes until the process table is full or you hit the
> per-user process limit, at which point fork() will begin to fail,
> which will then most likely terminate the calling process (unless you
> trap the error).

No sorry, this is definitely not the _full_ code ... it just forks
under some condition (usually, there are never more than 3 or 4
children) ... the point is that the father should never die, but it
does !

> I'd be curious to know how this could have worked earlier, or how you
> could expect it to behave differently.

It does, believe me ;-)

Martin

-- 
Martin Preishuber - Student, ECLiPt Core Member, SysAdmin
http://eclipt.uni-klu.ac.at,
mailto:Martin.Preishuber at stuco.uni-klu.ac.at

I respect faith, but doubt is what gives you an education.
-- Wilson Mizner