[Shorewall-users] NFS mount fail although NFS trafic go thru

[Reuben D. B. <tec...@vo...>]

Hi,
I find this really curious. I'm not real sure what causes this problem, b=
ut I=20
thought someone else might have encountered this.=20
Let's say I have a machine "prod" that runs shorewall (1 IP 1 interface=20
(eth0)) using the Quick Start guides. I open the ports to let NFS traffic=
s go=20
thru, just as described in www.shorewall.net.

So, prod is the NFS server. Now when I try to mount nan NFS exported dir =
from=20
another machine, mount will fail (RPC time out). In the /var/log/messages=
 of=20
prod, I find this:

Feb 18 11:18:14 prod kernel: Shorewall:net2all:DROP:IN=3Deth0 OUT=3D=20
MAC=3D00:04:75:ab:e7:26:00:10:dc:27:e3:d7:08:00 SRC=3D160.36.28.203 DST=3D=
<prod_ip>=20
LEN=3D172 TOS=3D0x00 PREC=3D0x00 TTL=3D64 ID=3D0 DF PROTO=3DUDP SPT=3D659=
 DPT=3D942 LEN=3D152=20

I'm curios why my "mount" command even try to access port 942.=20
But if I do 'shorewall stop' and 'shorewall clear' first, and then mount =
the=20
NFS export from another machine, and then bring shorewall up with 'shorew=
all=20
start', everything is OK. NFS traffics can go thru fine.=20

So why the mount command try to use different ports that what's specify? =
Is=20
this the OS problem (prod is an Redhat Enterprise 3, while the other is a=
 RH=20
7.3) ?

I vaguely remember then the DPT is not always the same, which even makes =
this=20
weirder.

RDB =20
--=20
Reuben D. Budiardja
Department of Physics and Astronomy
The University of Tennessee, Knoxville, TN
---------------------------------------------------------
"To be a nemesis, you have to actively try to destroy=20
something, don't you? Really, I'm not out to destroy=20
Microsoft. That will just be a completely unintentional=20
side effect."
                 - Linus Torvalds -