Mike Leong wrote:
> Hi,
>
> I'm currently using squid to reverse proxy cgi requests to our
> backend. I'm currently using some really aggressive refresh_pattern
> rules to make sure cache objects never get automatically purged. If
> something got changed in our backend servers, I manually purge the
> affected objects.
>
>
> As the number of objects being cached increase, it's getting more
> difficult to purge stuff. We're prepare to fix our backend webservers
> to do whatever it takes to make it such that squid can fetch new fresh
> objects, should anything in the backend get updated.
>
> My question is, what can we do to our backend to accomplish this.
>
> Currently, here's a sample header, dumped from a cache object:
>
> HTTP/1.1 200 OK
> Set-Cookie: blabla Expires=Sun, 08-Jul-2007 08:53:52 GMT #1 year from
> request
> TokenId: token-5102989651042811735-115234883242417937
> Content-Type: image/jpeg
> Content-Length: 18765
> Date: Sat, 08 Jul 2006 08:53:52 GMT
> Server: Apache-Coyote/1.1
> Connection: Keep-Alive
>
> We can probably add any other http header related to cache control, w/
> the exception of expiration date, since we have no idea when new data
> will be modified in the backend.
>
> thx
>
> mike
>
http://www.web-caching.com/mnot_tutorial/tips.html#SCRIPT
The whole tutorial is a good read, but this section deals specifically
with writing cacheable scripts.
Chris
Received on Thu Nov 30 2006 - 17:48:14 MST
This archive was generated by hypermail pre-2.1.9 : Fri Dec 01 2006 - 12:00:03 MST