[lkml]   [1999]   [Feb]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: lockd does not work as expected
On Sun, 28 Feb 1999, Steven N. Hirsch wrote:
> On Sat, 27 Feb 1999, Mike Harrelson wrote:
> > On Sat, 27 Feb 1999, Steven N. Hirsch wrote:
> > > > Thanks a lot. I'll try to replace the flock system
> > > > call by any of the others. Actually flock worked
> > > > perfectly as long as I just use solaris nodes. So
> > > > the linux system is not "consistent with existing
> > > > practice on other UNIX's".
> > >
> > > Yes, the semantics are not consistant across platforms. AIX and Digital
> > > Unix 4.x also seem to support flock() over NFS. Which Unices _don't_
> > > support it (just for curiousity)?
> >
> > Digital Unix, as of 4.0D, did not support NFS locking via flock(). It was
> > a local lock only.
> Odd. The last time I looked into this, it acted as if it was locking over
> NFS. I ran one little Perl program on the Alpha, and a second on an AIX
> NFS client machine. It certainly acted as if they were respecting each
> other's flock(). And, yes, I checked the Perl sources. It's definitely
> using flock().

I'd check again using strace or such. It definitely doesn't work here over
NFS, just locally. We tested programs (written in C) in a mixed environment
of Solaris and DU boxes. fcntl() locks were recognized by other machines,
flock() locks were not. There was, however, a newsgroup posting by a DEC
employee some months back stating that they may extend flock() to work over
NFS in 4.0E. I haven't verified if they did or not. I hope not.

-- mikeh

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
Please read the FAQ at

 \ /
  Last update: 2005-03-22 13:50    [W:0.043 / U:1.308 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site