Re: [squid-users] Multiple Accelerators

From: Brian <hiryuu@dont-contact.us>
Date: Thu, 30 May 2002 13:06:15 -0400

On Thursday 30 May 2002 11:01 am, AJ Lemke wrote:
> Hello Squid Group,
>
> My company is using squid in an website acceleration mode and we would
> like to add more squid servers to help with load and redundancy. I am
> looking for information pertaining to the sharing of information between
> squid boxes ie. can squid A give information to squid B? If so where
> can I find some documentation on this?

The FAQ has all kinds of information on cache peers (in this case, sibling
caches). I found it more efficient to run all of the accels
independently, simply retrieving misses from the origin servers, but YMMV.

> Another question. Is there a way to force squid to serve out expired
> pages if an error occurs on the backend server? An example, massive
> hardware failure on a web server and squid is recieving errors.

Try turning down connect_timeout. Once it times out, it should return the
stale file. It does with the origin servers set as parent caches, anyway.

        -- Brian
Received on Thu May 30 2002 - 11:06:18 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:08:17 MST