You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Rasmus Lerdorf <ra...@lerdorf.on.ca> on 1998/04/14 05:22:58 UTC

catching setgid error?

A couple of times now I have misconfigured my httpd.conf such that the
Group directive pointed to a bogus group.  When I start httpd it doesn't
complain, it just happily spews an endless stream of:

[Tue Apr 14 05:18:27 1998] [alert] (22)Invalid argument: setgid: unable to
set group id to Group 4294967295
[Tue Apr 14 05:18:27 1998] [alert] (22)Invalid argument: setgid: unable to
set group id to Group 4294967295
[Tue Apr 14 05:18:27 1998] [alert] (22)Invalid argument: setgid: unable to
set group id to Group 4294967295
[Tue Apr 14 05:18:27 1998] [alert] (22)Invalid argument: setgid: unable to
set group id to Group 4294967295
[Tue Apr 14 05:18:27 1998] [alert] (22)Invalid argument: setgid: unable to
set group id to Group 4294967295
[Tue Apr 14 05:18:27 1998] [alert] (22)Invalid argument: setgid: unable to
set group id to Group 4294967295

to the error_log file, and the server is utterly useless.  Would it not be
an idea to have a setgid error be a fatal error?  

-Rasmus


Re: catching setgid error?

Posted by Marc Slemko <ma...@worldgate.com>.
On Mon, 13 Apr 1998, Dean Gaudet wrote:

> In 1.2 it was a fatal error.  This bug was introduced when I fixed the
> "log files are created with the wrong group" bug by moving the setgid to
> the child.
> 
> But I'm not sure how portable it would be to do a test setgid in the
> parent and then setgid back...
> 
> A more portable solution, and more general, would be for a special child
> exit code (say, 42 for example) that means to the parent "abort everything
> immediately, something is fatally wrong".  There's a few other errors
> which are of similar caliber.
> 

The other interesting (but unrelated) child exit codes would be ones for
doing a restart (ie. HUP), graceful restart, "normal" exit, etc. 



Re: catching setgid error?

Posted by Dean Gaudet <dg...@arctic.org>.
In 1.2 it was a fatal error.  This bug was introduced when I fixed the
"log files are created with the wrong group" bug by moving the setgid to
the child.

But I'm not sure how portable it would be to do a test setgid in the
parent and then setgid back...

A more portable solution, and more general, would be for a special child
exit code (say, 42 for example) that means to the parent "abort everything
immediately, something is fatally wrong".  There's a few other errors
which are of similar caliber.

Dean

On Mon, 13 Apr 1998, Rasmus Lerdorf wrote:

> A couple of times now I have misconfigured my httpd.conf such that the
> Group directive pointed to a bogus group.  When I start httpd it doesn't
> complain, it just happily spews an endless stream of:
> 
> [Tue Apr 14 05:18:27 1998] [alert] (22)Invalid argument: setgid: unable to
> set group id to Group 4294967295
> [Tue Apr 14 05:18:27 1998] [alert] (22)Invalid argument: setgid: unable to
> set group id to Group 4294967295
> [Tue Apr 14 05:18:27 1998] [alert] (22)Invalid argument: setgid: unable to
> set group id to Group 4294967295
> [Tue Apr 14 05:18:27 1998] [alert] (22)Invalid argument: setgid: unable to
> set group id to Group 4294967295
> [Tue Apr 14 05:18:27 1998] [alert] (22)Invalid argument: setgid: unable to
> set group id to Group 4294967295
> [Tue Apr 14 05:18:27 1998] [alert] (22)Invalid argument: setgid: unable to
> set group id to Group 4294967295
> 
> to the error_log file, and the server is utterly useless.  Would it not be
> an idea to have a setgid error be a fatal error?  
> 
> -Rasmus
> 
>