Topic: New httpd version has problems with mod_status.so

[holck]

In my log file /var/log/httpd/admin_error_log I now see this message

Code: [Select]

Syntax error on line 29 of /etc/httpd/admin-conf/httpd.conf: Cannot load /etc/httpd/modules/mod_status.so into server: /etc/httpd/modules/mod_status.so: undefined symbol: ap_copy_scoreboard_worker
The problem is described here:
http://mattiasgeniar.be/2014/07/28/httpd-cannot-load-mod_status-so-into-server-undefined-symbol-ap_copy_scoreboard_worker/

For me, both the administrative and the ordinary web seems to be running ok.

My httpd-version is httpd-2.2.3-87.el5.centos

[Daniel B.]

Strange, I've updated a lot of server last week without this issue. Anyway, please, open a bug so it can be tracked

[CharlieBrady]

In my log file /var/log/httpd/admin_error_log I now see this message

I believe that is an indication that your old httpd binary is/way still running after the httpd update - and it was responding to a HUP signal (i.e. an apache graceful restart). Once a full restart of apache has completed (e.g. during reconfiguration/reboot) you won't see that message again.

I presume that you have already done the reconfiguration reboot. So now do:

sv h /service/httpd-admin

and check your httpd/admin_error_log file. You won't see that warning re mod_status.so. Correct?

[holck]

Correct

Code: [Select]

# sv h /service/httpd-admin
# tail /var/log/httpd/admin_error_log
[Tue Jul 29 01:12:03 2014] [notice] Digest: generating secret for digest authentication ...
[Tue Jul 29 01:12:03 2014] [notice] Digest: done
[Tue Jul 29 01:12:03 2014] [warn] pid file /var/run/httpd-admin.pid overwritten -- Unclean shutdown of previous Apache run?
[Tue Jul 29 01:12:03 2014] [notice] Apache configured -- resuming normal operations
[Thu Jul 31 20:57:42 2014] [notice] SIGHUP received.  Attempting to restart
[Thu Jul 31 20:57:42 2014] [notice] Digest: generating secret for digest authentication ...
[Thu Jul 31 20:57:42 2014] [notice] Digest: done
[Thu Jul 31 20:57:42 2014] [notice] Apache configured -- resuming normal operations

So what's happening is:

• Update of httpd, no restart
• Automatic attempt to graceful restart old httpd because of weekly log rotation
• Old httpd dies, leaving the error in the log file
• Because of sv, a new httpd is started, causing the warning on the overwritten pid file in the log
• This succeeds fine - no problem

So because of the automatic restart of httpd, it's mostly just log noise