From: Jim Jagielski <jim@jaguNET.com>
Message-Id: <199606091553.LAA06844@shado.jaguNET.com>
Subject: Logging and syslogd
To: new-httpd@hyperreal.com
Date: Sun, 9 Jun 1996 11:53:51 -0400 (EDT)
Content-Type: text
Sender: owner-new-httpd@apache.org
Precedence: bulk
Reply-To: new-httpd@hyperreal.com

Although it's true that most systems (and implementations of syslogd)
cannot support the load if access_log for busy sites, error_log is
written to MUCH less frequently. In that way, using syslog to log
error messages makes much more sense. That way, we can openlog/syslog/
closelog wrap the error logging saving a fd most of the time.

However, having Apache jump through all these hoops just to save fd's
seems a waste of our time, IMO. Look at INND, they just say "if you run
out of fd's, you can look into sfio." In other words, it's really not
_our_ problem, but rather a limitation of your OS. Here's a way to get
around it. Do we really want to work so hard so that systems limited
to 64 or 128 fd's can run a hundred v-hosts? I would guess that they would
hit other limits as well, as far as TCP/IP-stack design, performance,
etc...

To me, the question of internal-logging vs syslog() is one of which
may be better for users... We may save a few fd's here and there, but
it doesn't come for free.
-- 
Jim Jagielski  << jim@jaguNET.com >>   |      "That's a Smith & Wesson,
  **  jaguNET Access Services  **      |       and you've had your six" 
      Email: info@jaguNET.com          |             - James Bond
++    http://www.jaguNET.com/         +++      Voice/Fax: 410-931-3157       ++