More bizarre behavior! was Re: bizarre lag in NCSA's httpd 1.2? (Brian Behlendorf)
Date: Thu, 21 Apr 1994 23:13:45 +0200
Message-id: <9404212114.AA07140@barney>
Precedence: bulk
From: (Brian Behlendorf)
To: Multiple recipients of list <>
Subject: More bizarre behavior! was Re: bizarre lag in NCSA's httpd 1.2?
X-Listprocessor-Version: 6.0c -- ListProcessor by Anastasios Kotsikonas
Organization: WIRED Ventures
Organization: WIRED Ventures
In article <2p4gi7$>,
Rob McCool <> wrote:
>How fast is your DNS? httpd 1.2 does twice the number of resolutions it used
>to, which will be #ifdefed in 1.21. You should take a look at how many mbufs
>are dedicated to networking in your kernel, too, since the system may also
>be thrashing in the kernel (too many TCP connections to handle to port 80).

Actually, a look through error_log (duh! should have done this before)
shows this:

[Thu Apr 21 13:24:23 1994] httpd: caught SIGSEGV, dumping core
[Thu Apr 21 13:25:14 1994] httpd: caught SIGSEGV, dumping core

There are tons of these.  At first I thought this might be the
culprit, but in testing this theory out, I found it didn't always
correlate to those lags.  However, it DID correlate to someone
accessing our site via Lynx (2.2) when looking at a page which uses
the server-side includes, which we are using heavily here for
copyright notices.  The document still gets delivered to Lynx, only
without the copyright notice (which is a very bad thing).  I also
couldn't find a core file it said it was producing.

This does sound like two distinct bugs, but it wouldnm't surprise
me if they were related.

Apologies if this was already discussed and fixed... but if not, this
seems pretty serious.