On Thu, May 26, 2005 at 05:06:02PM -0700, Christian Seberino wrote:
> Lan
> 
> Thanks!  What is async and why fixes NFS problems like slowness?
> I use Autofs to NFS mount directories.  Does this async
> gotcha still apply?
> 
> Chris
> 

'K, I'm doing this from memory. For real information, goto the Linux
Documentation Project.

There are two kinds of NFS mount, hard and soft. Hard mounts demand a
response ... they wait on the connection. Soft mounts send something and
then go on to do other business ... they'll attend to the return if and
when it comes back. This is actually a fundamental configuration to any
socket connection.

I consider hard mounts in NFS to be an anachronism to when networks
couldn't be trusted to pass information, and critical systems simply had
to wait for the response. There may be uses today in other socket
connections, like requests that have to be serviced before the requester
can continue, but in NFS I don't see the point. Go soft all the time is
my attitude.

The reason a hard mount can cause slowness is that the server may have
other things it's doing, and the response to  your request may be
delayed. If so, your box is going to wait on the return before it does
anything else. You'll see this as slow response.

I don't know how Autofs mounts default, and obviously I'm postulating
here that this *may* be at the root of your problem. NFS works great for
most home installations, and if it's going slow for you, my first idea
is to check/change your configurations. 

So, going back to my last post, I'd suggest that you change the mount
line in your client's /etc/fstab along the lines of the example I posted
and see if that solves the problem. If yes, cool -- we all learned
something. If no, bummer -- post back and we'll do more thinking.

Good luck,

-- 
Lan Barnes                    [EMAIL PROTECTED]
Linux Guy, SCM Specialist     858-354-0616


-- 
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to