Re: Making metadata storage SQL-driven

[Jamie McCracken <jamiemcc blueyonder co uk>]

Alexander Larsson wrote:
> On Thu, 2005-09-01 at 13:47 +0100, Jamie McCracken wrote:
>
> > Alexander Larsson wrote:
>
>
> > > Requiring such a setup is pretty demanding, for the conceptually trivial
> > > thing of having two personal desktop machines. We all know this won't
> > > happen at most places.
> >
> > Well the only other thing I can think of is to use external lock files 
> > which are opened with O_CREAT and O_EXCL to serialise access in a 
> > multisession environment. Will that be atomic on NFS?
>
>
> Locking over NFS is handled by a separate protocol and daemon (lockd).
> We tried to use that in gconf, but apparently NFS locking is just
> totally broken on too many systems for it to work. 

yes though how does gconf solve this (or is it single session only)?

I also read the following as a work around on NFS:

"O_EXCL is broken on NFS file systems, programs which rely on it for 
performing locking tasks will contain a race condition. The solution for 
performing atomic file locking using a lockfile is to create a unique 
file on the same fs (e.g., incorporating hostname and pid), use link(2) 
to make a link to the lockfile. If link() returns 0, the lock is 
successful. Otherwise, use stat(2) on the unique file to check if its 
link count has increased to 2, in which case the lock is also successful."


--
Mr Jamie McCracken
http://www.advogato.org/person/jamiemcc/