[squid-users] acl and cache

From: Aurélien Bras <aurelien.bras@dont-contact.us>
Date: Thu, 4 May 2006 11:42:32 +0200

---------- Forwarded message ----------
From: Aurélien Bras <aurelien.bras@gmail.com>
Date: 4 mai 2006 11:18
Subject: Re: [squid-users] acl and cache
To: Henrik Nordstrom <henrik@henriknordstrom.net>

2006/5/3, Henrik Nordstrom <henrik@henriknordstrom.net>:
> ons 2006-05-03 klockan 14:38 +0200 skrev Aurélien Bras:
>
> > Oh I understand it's only a header, this is why there is text/html, so
> > it's work fine :D.
>
> Well, there shouldn't be text/html on an gif image. This indicates the
> web server is slightly broken..
>
Strange, my first request give :

127.0.0.1 TCP_MISS/200 1466 GET $URL - DIRECT/216.239.59.103 image/gif
127.0.0.1 TCP_MISS/200 1546 GET $URL - FIRST_UP_PARENT/127.0.0.1 image/gif

And the second give :

127.0.0.1 TCP_MISS/304 232 GET $URL - DIRECT/216.239.59.103 text/html
127.0.0.1 TCP_REFRESH_HIT/200 1544 GET $URL -
FIRST_UP_PARENT/127.0.0.1 text/html

Indeed, with another file I have the same mime type for the first
request (MISS) and for the second (TCP_REFRESH_HIT). The gif should be
broken, you are right.

Just to be sure, I have made test on one file which give one time a
refresh_hit and one time a refresh_miss.

with TCP_REFRESH_HIT, I have an average of 400ms
with TCP_REFRESH_MISS, I have an average of 800ms

My average ping on the server is 190ms.

I have a SWAPOUT entry only for TCP_REFRESH_MISS. So I think
TCP_REFRESH_HIT use the cache.

Is it normal to have only 400ms between a hit and a miss, am I on the
good way with my test ?

Thanks.

Aurélien.
Received on Thu May 04 2006 - 03:42:34 MDT

This archive was generated by hypermail pre-2.1.9 : Thu Jun 01 2006 - 12:00:01 MDT