Possible use for memcache?

[mike]

Ah, last I read it was something like "I'd like to port it fully to
FUSE but haven't finished it"

I would but I wouldn't implement it properly. :) If it was PHP-based,
sign me up. C? I couldn't write it to save my life. Especially if I
wanted something clean.

There might be some concerns of mine that I suppose don't need to be
addressed any further in this email thread or anything. You can ignore
me. I need to retire back to my cave here. I've never seen that clean
of HTTP requests, but I guess MogileFS only talks the headers it
needs, not like a normal thicker HTTP server.

Thanks.

- mike

On 11/28/05, Brad Fitzpatrick <brad at danga.com> wrote:
> I did make it work with FUSE, a long time ago.  It's bitrot and I lost it.
> Do it yourself.. it's easy.
>
> But I think you're confused about how "heavy" HTTP is.  Look:
>
>   GET /some-file HTTP/1.0\r\n
>   \r\n
>
> Holy shit!  That's it!
>
> Then you get the data sendfile(2)'d back to you after this huge header:
>
>  HTTP/1.1 200 OK
>  Content-Length: 23983
>
>  <data>
>
> Man, that's heavy.
>
> - Brad
>
>
> On Mon, 28 Nov 2005, mike wrote:
>
> > You should work on porting MogileFS to FUSE finally, so it doesn't
> > have to be coded into the specific application and would just be a
> > transparent filesystem to everything else :) Maybe cut down using HTTP
> > and make it a quicker, proprietary daemon like memcached, to save the
> > protocol overhead?
> >
> > </random ideas>
> >
> > - mike
> >
> > On 11/28/05, Brad Fitzpatrick <brad at danga.com> wrote:
> > > For a cache, it's fine.
> > >
> > > I wouldn't trust its failover modes for a filesystem.
> > >
> > >
> > > On Mon, 28 Nov 2005, mike wrote:
> > >
> > > > Brad! Are you telling me memcached isn't perfect?!?!?!
> > > >
> > > > :)
> > > >
> > > > On 11/28/05, Brad Fitzpatrick <brad at danga.com> wrote:
> > > > > Then you haven't seen all the failure cases.  :-)
> > > > >
> > > > > - Brad
> > > >
> > > >
> > >
> >
> >
>