From: mark edwards <medwards@4degrees.cc>
Subject: nfs mounts on linux
   Date: Mon, 20 Jun 2005 15:22:17 -0700

hey folks,

regards, mark

On Mon, 2005-06-20 at 15:22 -0700, mark edwards wrote:
> [posted to rush.general]
> 
> hey folks,
> 
> what is the best way to automount an nfs file server from a linux 
> workstation? The server is running OSX.3 and the workstation is SUSE 
> 9.3. 

I would do it by adding an entry into /etc/fstab and have an option for
auto

Something like:

server:/Volumes/share   /media/share nfs   (options) 0 0

There are lots of options that you can tweak for your needs.

Make sure that /mnt/share is there, if not make it. Then run mount -a


> Also does anyone know the speed comparisons with smb and nfs with 
> the same setup? 

In my experience OSX is slightly slower in NFS on 10.3. The first hit
was always slower till it cached anyway. AFP was the fastest. SMB
second,\ although that does spawn a child process for ever connection.
AFP doesn't thread across the Xserves procs.

I think it depends how many users are hitting the server. Less than 30
for 3D on SMB/NFS and you should be ok.

Dylan


> Thanks...
> 
> regards, mark

Dylan Penhale wrote:

[posted to rush.general]



On Mon, 2005-06-20 at 15:22 -0700, mark edwards wrote:

[posted to rush.general]

hey folks,

what is the best way to automount an nfs file server from a linux workstation? The server is running OSX.3 and the workstation is SUSE 9.3.

I would do it by adding an entry into /etc/fstab and have an option for
auto

Something like:

server:/Volumes/share   /media/share nfs   (options) 0 0

There are lots of options that you can tweak for your needs.

Make sure that /mnt/share is there, if not make it. Then run mount -a

Also does anyone know the speed comparisons with smb and nfs with the same setup?

In my experience OSX is slightly slower in NFS on 10.3. The first hit
was always slower till it cached anyway. AFP was the fastest. SMB
second,\ although that does spawn a child process for ever connection.
AFP doesn't thread across the Xserves procs.

I think it depends how many users are hitting the server. Less than 30
for 3D on SMB/NFS and you should be ok.

	Someone also mentioned you should be sure to disable 'encryption'
	over AFP, so that your client/server doesn't spend any time doing
	that..!


--
Greg Ercolano, erco@(email surpressed)
Rush Render Queue, http://seriss.com/rush/
Tel: (Tel# suppressed)
Cel: (Tel# suppressed)
Fax: (Tel# suppressed)

Greg Ercolano wrote:

Something like:

server:/Volumes/share   /media/share nfs   (options) 0 0

	BTW, I'd advise using hard mounts as opposed to soft mounts.
	ie. for (options):

		bg,intr,hard,nfsvers=3

	..and if the nfsvers=3 doesn't work, just lop it off.
	But bg,intr,hard are all definitely recommended.

	'hard' will keep trying if the nfs server goes down.

	'soft' will not, and will cause intermittant mount timeouts
	and weird render problems (drop outs).

--
Greg Ercolano, erco@(email surpressed)
Rush Render Queue, http://seriss.com/rush/
Tel: (Tel# suppressed)
Cel: (Tel# suppressed)
Fax: (Tel# suppressed)

thanks Greg and Dylan, I had it right except for a error in spelling.

I had...

server:/Volumes/share   /media/share nfs  bg,imtr,hard 0 0

intr works much better than imtr

regards, mark

____________________________

mark edwards
web designer / animator
4degrees.cc


On Jun 20, 2005, at 8:25 PM, Greg Ercolano wrote:

[posted to rush.general]

Greg Ercolano wrote:

Something like:

server:/Volumes/share   /media/share nfs   (options) 0 0

	BTW, I'd advise using hard mounts as opposed to soft mounts.
	ie. for (options):

		bg,intr,hard,nfsvers=3

	..and if the nfsvers=3 doesn't work, just lop it off.
	But bg,intr,hard are all definitely recommended.

	'hard' will keep trying if the nfs server goes down.

	'soft' will not, and will cause intermittant mount timeouts
	and weird render problems (drop outs).

--
Greg Ercolano, erco@(email surpressed)
Rush Render Queue, http://seriss.com/rush/
Tel: (Tel# suppressed)
Cel: (Tel# suppressed)
Fax: (Tel# suppressed)