On Fri, 2002-05-24 at 01:38, Kent, Ian I. wrote:
> Colin Simpson wrote:
> 
> >
> > I think autofs is a vital project for Linux to be taken seriously in the
> > enterprise environment. I think for this to happen v4 (if not version at
> > least the funcationality) at a minimum is needed, and to really compete
> > with commercial Unix all the features on the TODO list will be required.
> >
> 
> Specifically which features in v4 and not in v3 are you talking about.
> 
> 
The /net feature is the big one for me. 

Here rather than messing with symlinks to simulate this, we just run
autofs for our standard maps and use amd for /net. This doesn't seem to
cause a problem.

I think I've posted this before, but if it's of any interest to anyone
here is is again. In fact the standard Red Hat amd config only has an
entry to /net. I altered their amd.net file to allow the local machine
to mount without using nfs. Here's our amd.net.

/defaults fs:=${autodir}/${rhost}/root/${rfs};opts:=nosuid,nodev
*       host!=${key};rhost:=${key};type:=host;rfs:=/ \
        host==${key};type:=link;fs:=/

It would be nice to ditch amd and just use autofs.

I don't want to sound complaining (I know it is far from trivial to do)
I would also like direct maps and the ability to see automount points
before they are mounted (as Solaris does).

If this involves starting again with modifying v3 in a way considered
less dangerous with no races and jumping to v5. So be it. As long as the
project can move forward. These missing features make Linux look less
credible against the commercial Unicies. Unfair, but it's a rod to beat
us with. And the inevitable, if it can't even do this "simple" thing
from various people.

Thanks

Colin


Reply via email to