Re: [squid-users] http partial caching

From: Henrik Nordstrom <[email protected]>
Date: Wed, 23 Mar 2005 09:41:21 +0100 (CET)

On Wed, 23 Mar 2005, Jun OKAJIMA wrote:

> case A. You are disconnected in the first 30MB. Then download the rest
> with resuming. In this case, squid can cache a full 100MB file
> at the end? and of course, the next users can benefit from it?

What happens here depends on your quick_abort settings in squid.conf, and
how the abort happened.

If the connection between Squid <-> Origin server gets aborted then the
content is discarded.

> case B. While you are downloading the rest 70MB with resuming, If another
> user is going to download the first 30MB, what happens? The user
> can benefit from the 30MB cache?

As above.

Squid does not cache partial content (yet). As a result if there is only a
partial content downloaded and the connection to the origin server gets
closed for one reason or another the so far downloaded content is
discarded.

To mitigate some of this Squid has a feature to allow continuing
downloading content even when the client has aborted. How aggressive Squid
does this is controlled by the quick_abort parameters in squid.conf.

I normally recommend to have these set to their strictest settings,
forcing Squid to abort downloads immediately when the client aborts. I
find there is too many issues with bandwidth wasting if this continuation
of donloads is allowed.

Regards
Henrik
Received on Wed Mar 23 2005 - 02:34:12 MST

This archive was generated by hypermail pre-2.1.9 : Fri Apr 01 2005 - 12:00:02 MST