On Thu, 27 Nov 2003, Sander Temme wrote:

> >>> some reason, Apache keeps SEGFAULT'ing every 4 - 5 days. When apache
> >>> SEGFAULTS, it doesn't produce a core file, and spits out the following
> >>> items in our error_log:
> >>
> >> Did you set your ulimit -S -c to a value other than 0? You need to do that
> >
> > I sure did. I read somewhere that the Linux 2.4 kernel won't allow
> > processes that change their user ids to dump core. I see why they did
> > that, but it is really putting me in a bind. Are there any other ways
> > to determine why Apache is dieing? ANy other thoughts on why the
> > prtctl plugin isn't allowing Apache to dump core?
>
> Yeah, that's strange because, looking at the code, this is exactly what the
> module is supposed to do. You could run your server as non-root, so it
> doesn't setuid and setgid on the children: this would take away the kernel's

That is a good idea, but there has to be a way to debug things in their
current state. Since this is the configuration that causes Apache to
SEGFAULT, I would like to capture a core/stack trace when this
configuration goes south. I am hoping one of the developers can shed some
light on this.

> issues with letting the children dump core. The server would have to be able
> to bind to its listening port(s) (use >1024) and write to its log directory
> and files.
>
> S.
>
> --
> Covalent Technologies                 [EMAIL PROTECTED]
> Engineering group                    Voice: (415) 856 4214
> 303 Second Street #375 South           Fax: (415) 856 4210
> San Francisco CA 94107
>
> PGP FP: 7A8D B189 E871 80CB 9521  9320 C11E 7B47 964F 31D9
>
> =======================================================
> This email message is for the sole use of the intended
> recipient(s) and may contain confidential and privileged
> information. Any unauthorized review, use, disclosure or
> distribution is prohibited.  If you are not the intended
> recipient, please contact the sender by reply email and
> destroy all copies of the original message
> =======================================================
>

Reply via email to