On sön, 2008-06-29 at 01:24 +0800, WestWind wrote:
> Thanks for henrik's advice, I am running a file server that often
> requested by multi-thread download tools, if squid can prefetch and
> cache files, It will make my back-end server do less work.
>
> I find some email lists talking about this matter here:
> http://www1.uk.squid-cache.org/mail-archive/squid-users/200404/0962.html
> it's replied by henrik too, but I am still not understand How squid
> think about "abort" ? squid treat all the Range request as "abort" ?
> What about the browser window is closed by some one while requesting
> a file?
quick_abort_min -1 KB
range_offset_limit -1 KB
The first disables quick_abort, making Squid continue "downloading" the
file even after the client has disconnected.
The second enables full range prefetching, making Squid always fetch the
complete file and then split it in ranges.
Squid does not differentiate between someone closing the window while
downloading from someone requesting just a part of the file. In both
cases it was a partial request, and if quick_abort is enabled (the
default) then Squid will stop downloading the object and discard the
result when no client is listening for the result.
Regards
Henrik
This archive was generated by hypermail 2.2.0 : Sun Jun 29 2008 - 12:00:06 MDT