[Linux-cluster] Re: Csnap instantiation and failover using libdlm

[Benjamin Marzinski]

On Tue, Oct 19, 2004 at 01:48:15PM -0400, Daniel Phillips wrote:
> On Tuesday 19 October 2004 11:20, Benjamin Marzinski wrote:
> 
> > O.k. stupid question time: If a userspace process graps this
> > exclusive lock, and then dies unexpectedly, does the lock
> > automatically get freed?
> 
> Yes.  Though I haven't looked closely at this, it seems the locks are 
> cleaned up when the fd that libdlm creates to pass lock completions to 
> userspace is closed.  It's not strictly coupled to process exit, it's 
> even more sensible.
> 

o.k. As long as a server never loses it's lock while there is the possibility
that it could write to disk, and the lock gets automatically freed once the
process has died, I don't have any objections.

-Ben