Re: [squid-users] Accelerator Problem

1. That's a LOT of cache directories. Try toning it down to 16 256 or so.
2. Are the reply codes from the origin server 304 (not modified) or 200?
3. Does this affect image files as well as pages?
4. Turn on store.log and see if squid is doing a SWAPOUT on objects
(saving them to disk)
5. Are you testing this with 'Reload' in your browser? This bypasses the
cache, although you can disable that in squid.
6. Use telnet to manually make a request to the origin server. Does the
response include last modified or expires headers? Do they include cache
control or vary headers?

        -- Brian

On Tuesday 30 October 2001 08:52 pm, ray.taft@xfrnetworks.com wrote:
> Could use some expert help on this one...
>
> I have installed squid on a front end server to accelerate a backend web
> server.
>
> Everything seems to be working normally, except the following:
>
> 1: The cache server ALWAYS goes back to the origin server for the
> content (tail -f the access log on the origin server confirms this on
> every hit)
>
> 2: Performance is S L O W.. due in part (assuming) that the cache server
> connects to the HTTP server, grabs the data, and then serves it.
>
> The intent is to have the objects on the webserver stored in memory /
> disk on the squid box. When the cache is hit, it serves them from memory
> or disk, faster than what the slower web server could do, and DOES NOT
> go back to the origin server each time for data.
>
> Attached is my squid.conf file.
>
> The Squid Box is a Solaris 8 REV 10, Sun Netra T1, 440Mhz, 256MB RAM, 2
> X 9GB (10,000 RPM 5.2Ms/seek). I have 6GB dedicated to squid cache.
>
> Any help would be appreciated.
>
> Ray
Received on Tue Oct 30 2001 - 19:51:00 MST