HTTP error 408 not logged in ErrorLog

On 2005-08-31, I submitted the following bug to ASF Bugzilla:

http://issues.apache.org/bugzilla/show_bug.cgi?id=36433

"The apache ErrorLog does not log 408 (request timeout) errors.
Instead, they appear in the normal TransferLog. Can someone please fix
this behaviour or give me a hint as to where in the code these
decisions are made? I have an app that relies on "[error]" conditions
appearing in the error log. Thanks."

So far, I've heard nothing back and the problem persists. Can anyone
one here offer me some assistance as far as how to change this so that
408 errors log where they should (in the ErrorLog, not the
TransferLog). Thanks in advance.

Lew Payne wrote:
> On 2005-08-31, I submitted the following bug to ASF Bugzilla:
>
> http://issues.apache.org/bugzilla/show_bug.cgi?id=36433

That references httpd-1.3, which has been obsolete for over
three and a half years and has very little developer interest.
You might get fixes there, but it's not very likely unless you
report a critical or security bug (which you didn't), or do
the work yourself (or hire someone to do it).

--
Nick Kew

Thanks for the answer, Nick. I haven't seen a good business case yet
for switching everything over from apache 1.3 to 2.x, especially since
only the prefork MPM can be used (in FreeBSD) if you're running PHP
and modules which utilize non thread-safe libraries. I imagine the
only thing I'll gain, for the most part, is memory bloat (our load is
split across 4 web drones, each rinning around 400 apache processes).

I guess I'll take a stab at the source, and see if I can figure out how
certain error codes are handled, and what determines where they
appear. If not, as you say, I'll end up having to hire someone.

Lew Payne wrote:
> Thanks for the answer, Nick. I haven't seen a good business case yet
> for switching everything over from apache 1.3 to 2.x,

Well, that's your choice, but ...

especially since
> only the prefork MPM can be used (in FreeBSD) if you're running PHP
> and modules which utilize non thread-safe libraries.

That makes it look like an ill-informed choice. Threaded MPMs are
but one of many reasons to upgrade, and certainly aren't what
attracted me to Apache 2.

I imagine the
> only thing I'll gain, for the most part, is memory bloat (our load is
> split across 4 web drones, each rinning around 400 apache processes).

On the other hand, you could see a large reduction in your load,
even with prefork.

> I guess I'll take a stab at the source, and see if I can figure out how
> certain error codes are handled, and what determines where they
> appear. If not, as you say, I'll end up having to hire someone.

Sounds like you might be making your own case to upgrade there!

--
Nick Kew