Re: How to tell that my cache is overloaded...

From: Oskar Pearson <>
Date: Mon, 30 Jun 1997 14:10:31 +0200

Stephane Bortzmeyer writes:
> On Friday 27 June 97, at 18 h 27, the keyboard of Nigel Metheringham
> <> wrote:
> > I'd like a list of things to watch so that I can tell when our cluster of
> > cache machines is running out of steam. What points, other than general
> > response time, which tends to depend very much one what you are looking
> It seems that response time is THE main parameter to watch since this is
> why we use caches in the first time! (To surf faster.)
agreed - it's no use if the cache is slowing you down more than 1 second on

> We monitor the response time for a file which is already in the cache
> (otherwise, you depend on things you do not control like the speed of the
> remote links). This is the main criterium of the load of the cache. We
what I do is run a we server on the same machine as the cache, so when
I do this test I simply query the web server, (twice, to get it out of
sleep mode) and then a couple of get's from the proxy, and figure
it out from there

cd /home/httpd/html/cache/vitals/vitals2/

echo "############" >> laglog
echo "cache:`/usr/local/bin/echoping-nocache -h / cache:80`" >> laglog
echo "cache-squid:`/usr/local/bin/echoping-nocache -h cache1:8080`" >> laglog

I have modified echoping so that it sends a "no-cache" and called this

newt:~/squid/echoping/echoping-nocache.src> diff -u http.c ../http.c
--- http.c Thu Apr 24 11:49:25 1997
+++ ../http.c Wed Oct 16 18:39:17 1996
@@ -9,7 +9,7 @@
   int size = 200; /* Enough? */
   char *sendline = (char *) malloc (size);
- sprintf (sendline, "GET %s HTTP/1.0\r\nPragma: no-cache\r\nUser-Agent: Echoping/%s\r\n\r\n",
+ sprintf (sendline, "GET %s HTTP/1.0\r\nUser-Agent: Echoping/%s\r\n\r\n",
           url, VERSION);
   return sendline;
Received on Mon Jun 30 1997 - 05:20:08 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:35:35 MST