Prev: patch status health check
Next: stmsboot on solaris 8 and 9
From: jay on 2 Apr 2010 13:31
solaris 10, u8 (oct 09)
x86

svcs -a | grep nfs
online 12:09:49 svc:/network/nfs/status:default
online 12:09:49 svc:/network/nfs/nlockmgr:default
online 12:09:51 svc:/network/nfs/mapid:default
online 12:09:51 svc:/network/nfs/cbd:default
online 12:09:52 svc:/network/nfs/rquota:default
online 12:09:52 svc:/network/nfs/client:default
online 12:09:53 svc:/network/nfs/server:default

/etc/dfs/dfstab:
share -F nfs -o ro=flawed-s,root=flawed-s /VOLUMES/space
share -F nfs -o ro,anon=0 /VOLUMES/solaris/sparc
share -F nfs -o ro,anon=0 /VOLUMES/solaris/i86pc
share -F nfs -o ro,anon=0 /Q

flawed-s is a netgroup, which does contain the machine jgs.

the machine jgs CAN nfs mount
/VOLUMES/solaris...
/Q
from itself, and it mounts via nfs from other machines via
automounter.

it cannot mount /VOLUMES/space, permission denied.
/VOLUMES is all one mount point. mode 755, root:other
before the VOLUMES slice is mounted, the underlying
dir is mode 755, root:other
/VOLUMES/solaris, /VOLUMES/space are root:root, 755.

if i unshareall, and then
jgs# share -F nfs -o ro=jgs.arlut.utexas.edu,anon=0 /VOLUMES/space
jgs# mount -F nfs -o ro jgs:/VOLUMES/space /mnt
nfs mount: mount: /mnt: Permission denied
so it shouldn't be the related to the netgroup.

/mnt is 755, root:sys

but /Q works

jgs# mount -F nfs jgs:/Q /mnt
jgs# ls /mnt
BustOpen howitry

so i'm at a complete loss. i can't imagine what's going on.

j.
From: John D Groenveld on 2 Apr 2010 16:15
In article <4b565030-d2ee-4236-ad1b-e3c3834439ec(a)o30g2000yqb.googlegroups.com>,
jay <gl(a)arlut.utexas.edu> wrote:
>jgs# share -F nfs -o ro=jgs.arlut.utexas.edu,anon=0 /VOLUMES/space

Just a WAG, how about share -o ro=jgs or share -o ro=localhost ?
Guessing name lookup issue.

John
groenveld(a)acm.org
From: hume.spamfilter on 2 Apr 2010 17:21
jay <gl(a)arlut.utexas.edu> wrote:
> jgs# share -F nfs -o ro=jgs.arlut.utexas.edu,anon=0 /VOLUMES/space
> jgs# mount -F nfs -o ro jgs:/VOLUMES/space /mnt

What happens if you use "localhost" instead?

Also, I don't think NFS mounting filesystems onto the same host is a good
idea. I believe there are situations where you can wedge the system.
Use lofs instead.

--
Brandon Hume - hume -> BOFH.Ca, http://WWW.BOFH.Ca/
From: ITguy on 2 Apr 2010 17:33
> >jgs# share -F nfs -o ro=jgs.arlut.utexas.edu,anon=0 /VOLUMES/space
>
> Just a WAG, how about share -o ro=jgs or share -o ro=localhost ?
> Guessing name lookup issue.

To rule out the possibility of a name service error:
# share -F nfs -o ro=<IP_Address>,anon=0 /VOLUMES/space

To resolve the name resolution issue: On the NFS server, use the
command "getent hosts <client_IP>" and ensure that the first returned
result is the one present in the netgroup.

From: Stefaan A Eeckels on 2 Apr 2010 18:45
On Fri, 2 Apr 2010 21:21:33 +0000 (UTC)
hume.spamfilter(a)bofh.ca wrote:

>
> Also, I don't think NFS mounting filesystems onto the same host is a
> good idea. I believe there are situations where you can wedge the
> system. Use lofs instead.

Home directories are routinely automounted. There's no problem as long
as you don't put them in /etc/vfstab (in which case you can wedge a
machine at boot whatever the source of the filesystem).

--
Stefaan A Eeckels
--
"The most common of all follies is to believe passionately in
the palpably not true. It is the chief occupation of mankind."
--H. L. Mencken
 |  Next  |  Last
Pages: 1 2 3
Prev: patch status health check
Next: stmsboot on solaris 8 and 9