The new on-disk cache implementaion

[Juli Mallett]

Ahmed,

I went through several incomplete implementations that predate
Diego's, and I have plans to extend it beyond his work; I wanted to
start from a design that would extend to support the features and
functionality I intend to include, and some that were needed today,
including the ability to share a single on-disk cache between multiple
peers.

Upload and download both go into the cache, but they do not share
data, at least not yet.  So a segment from one peer will not be used
to deduplicate data from another peer.  Whether this is done in the
future is an open question; it raises a lot of issues about
configurations with many-to-many relationships.  It might be worth
having a configuration parameter to share a cache for local and remote
segments in one-to-one configurations.

Let me know if your issue persists with the latest code.

Thanks,
Juli.

On Fri, Apr 24, 2015 at 10:09 PM, Ahmed Al -Ghafri
<al-ghafri at hotmail.com> wrote:
> Hello Juli,
>
> Excellent and great advance in WANProxy for this month. Finally, on-disk
> cache is in progress
> to be supported officially. I wanted to ask, what is the difference between
> your on-disk cache implementation
> and Deigo implementation? I mean why you started from scratch, and not build
> on what Deigo has done?
>
> Another thing, in the current implementation, is the on-disk cache works two
> ways direction, I mean upload/download both are considered to fill the
> cache?
>
> BTW, last time I faced a problem showing error:[/zlib/inflate_pipe] ERR:
> virtual void InflatePipe::consume
> If you can help I would be appreciated; here is the link:
> http://lists.wanproxy.org/pipermail/wanproxy-wanproxy.org/2015-January/001555.html
>
> Regards,
> Ahmed