On Thu, 19 Dec 2002, David Douthitt wrote:
> Attached is a ptrace of a running process (which
> has 4 zombies). The zombies are process 7380,
> 7381, 7384, and 7385. As of sending time, these
> five processes (zombie and parent) still exist.
>
> David Douthitt
> CUNA & Affiliates
> UNIX Systems Administrator
> LPI Level 1, Linux+
> ddouthitt@xxxxxxxxx
>
>
> >>> Mark Goodwin <markgw@xxxxxxx> 12/18/02 7:50PM >>>
> On Wed, 18 Dec 2002, David Douthitt wrote:
>
> > pmie (used with pmie_check) is generating a lot of zombies.
> ...
>
> Thanks David, I've posted an (SGI internal) bug about this. In
> the mean time, if anyone has a fix please send it to the list.
David,
I've had no success trying to reproduce this, so I've added more
diagnostics in pmie to track what's going on with process creation and
harvesting.
What struck me as odd in your ptrace case was that the 4 forks were
immediately followed by 4 wait()s that returned 4 _different_ pids ...
so I can imagine a small number of zombies in a sort of convoy where
firing off a rule causes the zombies from the last rule firing to be
harvested ... but this in no way explains the number of zombies growing
unbounded.
If you get a chance, could you please download the latest PCP src
tarball from oss.sgi.com, and see if the problem is still present on
your system.
If it is, could you please try running pmie with -Dappl2 on the command
line, and send me the stderr from the pmie process.
|