Off-hand I'd say you could set up a page or cgi-script on the server
so that client A could fetch or post to that page/script, which would
then create a lock file on the server. Client A grabs the file, does
stuff, uploads it, whatever. When finished, client A posts/gets the
unlock page. In the mean time, any other client trying to access the
lock script/page gets a "busy signal" or spins until timing out.
Lather, rinse, repeat, until client A releases the lock, at which time
some other client finally gets access to the lock page.

All server-side, where the file resides. I think that's the only way
to go. The only curl involvement would be standard posting/getting
pages. The tricky stuff would be server side in whatever language you
care to use.

Ralph Mitchell

On 6/24/05, Haber, Orna <Orna.Haber_at_ca.com> wrote:
>
> The purpose behind the locking need is to avoid situation where more
> than one client is try to modify the same file (at the same time).
> Because of the file (sensitive) contents, it is important to wait until
> one client finish updating the file before another client starts
> updating it.
> What is the best way to provide this locking (since curl is not the
> answer)?
> Thanks,
> Orna
>
> -----Original Message-----
> From: curl-library-bounces_at_cool.haxx.se
> [mailto:curl-library-bounces_at_cool.haxx.se] On Behalf Of Dan Fandrich
> Sent: Friday, June 24, 2005 11:03 AM
> To: libcurl development
> Subject: Re: Locking a file on the Http Server
>
> On Fri, Jun 24, 2005 at 01:07:22PM -0400, Haber, Orna wrote:
> > Yes. You got this right.
> > My thought about using a shared object and provide a lock-function in
> > order to create a cookie on the server machine or create a
> > "lock"-key-value in the server's registry (DNS cache).
> > Of course I have the problem of using the same shared object by
> > processes running on different machines..(:-)
> >
> > -----Original Message-----
> > From: curl-library-bounces_at_cool.haxx.se
> > [mailto:curl-library-bounces_at_cool.haxx.se] On Behalf Of Ralph Mitchell
> > Sent: Thursday, June 23, 2005 8:07 PM
> > To: libcurl development
> > Subject: Re: Locking a file on the Http Server
> >
> > Let me see if I got this straight - there's a file on a web server,
> > and it's being access from multiple different clients. You want to
> > lock it so that it can only be accessed by one client at a time. Is
> > that right??
> >
> > If so, it sounds like something the web server needs to handle. The
> > curl instances on different machine won't have a clue about locks on
> > other machines.
>
> Is this really the requirement? I don't know what you're trying to do
> with
> this, but you must understand the potential for one requester to cause a
> denial of service for the others. If you just want to modify a file
> while
> clients are downloading, it can be easily done on a *NIX server without
> disrupting other clients from accessing the old version at the same
> time.
> If you're just trying to create a globally-unique value for each client,
> there are ways of doing that without blocking, too. In any case, as
> Ralph
> has said, none of this involves curl.
>
> >>> Dan
> --
> http://www.MoveAnnouncer.com The web change of address
> service
> Let webmasters know that your web site has moved
>
>
>
>
Received on 2005-06-24