[SUMMARY] httpd core dumps

From: Jean-Loup Risler <Jean-Loup.Risler_at_genetique.uvsq.fr>
Date: Mon, 18 Dec 95 16:16:34 +0100

Dear all,

Although I could compile NCSA httpd, it core dumped when I ran it.

The reason and the solution came from Christopher J. Grayce (thanks!). Here
it is:

>
> I had an interesting problem with httpd 1.5a on a 600 5/266 running
>DU 3.2c, which I just installed a week or so ago. In httpd.conf there
>is a part that reads:
>
>User nobody
>Group #-1
>
>in the distribution version (httpd.conf-dist). If I left this entry as
>is in my actual httpd.conf, or changed the number to any "negative"
>number, e.g. Group #-26 (26 is the group number I assigned to a "www"
>group) then the result was exactly as you reported --- httpd compiled
>fine, but when run crashed immediately with a memory fault. I found
>out by poking around with the source code that httpd was calling
>setgid with a negative group number, which is no good, and that there
>was a bug in the part of the code that was supposed to fail gracefully
>at that point, so that instead httpd would simply crash. The fix is
>simply to get rid of the minus sign, i.e. my http.conf now looks like
>
>User nobody
>Group #26
>
>and now I have no problem.
>
> I mailed a bug report to NCSA, and this is what I got back:
>...
> (stuff deleted)
>...
>
>The bug is in die, and the setting of &fakeit. Some OS's don't like
>negative gids, but some do. It is fixed in my internal version,
>but I'm looking for a couple other errors before releasing it.
>

Now I could launch httpd without any problem. Cheers!

Thank you also to Andy Sheaff, to Andrew C. Saylor who recommended using
the "apache" daemon available from www.apache.com, and to a guru whose mail
I deleted inadvertendly (sorry) who recommended using the CERN httpd
because it consumes less memory.

-- 
---------------------------------------------------------------------
| Jean-Loup Risler               |                                  |
| Universite de Versailles       | Tel: (33-1) 39 25 45 54          |
| Lab. Genome et Informatique    | Fax: (33-1) 39 25 45 69          |
| Batiment Buffon                |                                  |
| 45 Avenue des Etats-Unis       | email: risler_at_genetique.uvsq.fr  |
| 78035  Versailles Cedex France |                                  |
---------------------------------------------------------------------
Received on Mon Dec 18 1995 - 17:04:09 NZDT

This archive was generated by hypermail 2.4.0 : Wed Nov 08 2023 - 11:53:46 NZDT