Re: [squid-users] StoreID and concurrency?

On Wed, Jun 5, 2013 at 11:57 PM, Eliezer Croitoru <eliezer_at_ngtech.co.il> wrote:
> StoreID is not related directly to cache_dir so it's not suppose to affect
> dirty cache_dir.

StoreID uses the output from the helper as the key to store the cached
copy in cache_dir, does it not? So if the helper generates one key for
the video and a *different* key for the same video next time (if I
change the helper output), it will lead to inconsistency for that same
video, will it not? Particularly relevant to YT which splits a video
into parts (ranges), where each part gets its own key. So if I change
the key halfway while the video is playing, things get confusing.

BTW, with reference to your example:

#./new_helper.rb
http://freefr.dl.sourceforge.net/project/vlc/2.0.5/win32/vlc-2.0.5-win32.exe
OK store-id=http://dl.sourceforge.net.squid.internal/project/vlc/2.0.5/win32/vlc-2.0.5-win32.exe

Maybe a different example would demonstrate the usefulness of StoreID
a bit more. In your example above, the original URL is of a fixed
format which be cached by default by Squid. StoreID would not provide
any added value there. The URL-rewrite examples in StoreURL wiki would
be better in the sense that they actually show a re-write of an
un-cacheable URL into a cache-able one.

> if it works for you good for me and oyu.
> why do you want to reduce browser cache??
> it's not bad to have cache in the browser if you have enough ram.
>
> Eliezer

If you have a large browser cache, and you replay the video, you can't
tell for sure if the video came from the browser cache, or Squid
cache. Therefore one is liable to draw wrong conclusions as I did.
Received on Thu Jun 06 2013 - 13:25:51 MDT