apache13+mod_ssl portupgrade problem..

Paul Schmehl pauls at utdallas.edu
Thu Sep 22 10:00:01 PDT 2005


--On Thursday, September 22, 2005 18:12:44 +1000 Steve Monkhouse 
<stevem at protek.net.au> wrote:

>
>>> Hi guys..
>>>
>>> Ive been pulling my hair out for the last couple of hours over this to
>>> no avail...
>>>
>>> FreeBSD 5.2.1-RELEASE-p8
>>> P4 2.8
>>> 512mb RAM
>>>
>>>
>>> I 'had' a working apache13+modssl installation along with mod_perl.
>>> Everything was working perfectly until I decided yesterday to
>>> portupgrade the machine.
>>>
>>> I started off with perl being portupgraded via  'portupgrade -rR perl'
>>>
>>> Then I did the php modules via portupgrade php-*
>>>
>>> And anything else that I missed then got done as well..
>>>
>>> Everything now shows that its up-to-date eg..
>>>
>>> apache+mod_ssl-1.3.33+2.8.24_1      =   up-to-date with port
>>>
>>> however whenever I try to start apache with apachectl startssl it gives
>>> :
>>>
>>> /usr/local/sbin/apachectl startssl: httpd started
>>>
>>> but does not start.. nothing at all appears in the logs (error or
>>> access) even with debug enabled.. the only message that appears is in
>>> messages
>>>
>>> Sep 22 17:44:23 server kernel: pid 77689 (httpd), uid 0: exited on
>>> signal 11 (core dumped)
>>>
>>> Ive searched high and low for a solution, but to no avail..
>>>
>>> Ive deinstalled and reinstalled, deinstalled / install distclean etc etc
>>> but no diff..
>>>
>>> The sad thing is that this was working perfectly yesterday albeit with
>>> versions that haven't been updated for 2 months..
>>>
>>> What have I done wrong ? and how do I go about rectifying it ??
>>>
>> Try doing "apachectl configtest" to check your config files. I've seen
>> apache
>> just bail with no errors on a bad config.
>
> Unfortunately.. No..
>
> root at server:/usr/local/etc/apache# apachectl configtest
> Syntax OK
> Segmentation fault (core dumped)
>
> My config hasn't changed, and as I said was working perfectly before the
> portupgrade..
>
> Next idea ??

I had the exact same problem.  I discovered that, when you install perl 
from ports, it makes the port perl the active perl version.  (Read 
/usr/ports/UPDATING) If you try to switch to it (use.perl ports) it somehow 
switches back to the system perl and everything goes wacky and apache 
segfaults.  (For of habit I'll have to get out of now that I'm running the 
5 series OS.)

Try uninstalling and reinstalling perl from ports - then *do not* run 
use.perl ports - then type perl -v and you'll see that perl is now the 
ports version.

Then deinstall and reinstall apache.  That *should* solve the problem. 
(You may also have to reinstall some of your perl-dependent ports.)

One other thing.  If you just deinstall and reinstall apache (without doing 
the perl stuff first), you *should* see a warning during the apache install 
that perl could cause a segfault.  I missed it the first time.  :-(

I even installed apache 2.0.54 trying to solve the problem, but I could 
never get ssl to work, so I gave up and went back to apache13-mod-ssl. 
It's not your config.  Trust me. :-)

(Of course, type perl -v first.  If you're already running the ports 
version of perl, you shouldn't need to uninstall and reinstall it.)

Paul Schmehl (pauls at utdallas.edu)
Adjunct Information Security Officer
University of Texas at Dallas
AVIEN Founding Member
http://www.utdallas.edu/ir/security/


More information about the freebsd-questions mailing list