Slightly off topic NFS question

Slightly off topic NFS question

[Michael French]

    Don't ask why, but I have to get a Windows box and a Solaris box talking
over NFS.  The win box is the server and solaris box is the client.  I
installed the Services for Unix on the Win box and configured NFS (what
little there is to do).  I created a share and thought I was ready to rock.
I do a showmount -e on the Win box and the share shows up.  rpcinfo
servername shows all the processes I expect.
    From the Solaris client, I can do the same thing with rpcinfo and
showmount -e servername and get results back.  I had the firewall admin open
up all IP traffic between these to boxes (just for now, for testing, having
problems before this with just 2049 and 111 open).  I can mount the share
with no problem (mount -o rw -F nfs 192.168.0.10:/testmount /testmount).  It
mounts right away and running mount shows it with read/write/setuid on
server, but as soon as I try to write to the share, I get a "Stale NFS"
error.  I check the permissions on the mounted folder, changed to 777 before
I mounted, owned by nobody/nobody (I am root right now).  On the win server,
the log is showing a successfull mount, no errors.  The permissions on the
directory on the server are wide open for all users, including connecting
NFS clients.  Any idea what the hell could be going on?

I am using Solaris 8 and Win2K with SFU 2.2.  Win2K patched to SP3.

Thanks for any answers peope might be able to provide, I have to get this
working today.


Michael French

Re: Slightly off topic NFS question

[Paul Furness]

I'm afraid I can't give you a complete answer, just some pointers.
Unfortunately, I haven't used NFS server from a windows machine, so
these might be completely the wrong track. Anyway...

I have had problems getting samba to mount windows shares in the past,
and it seems likely that windows NFS uses some of the same auth
mechanism. The main problem I had was that, if you tried to mount the
windows share as a user that didn't exist _on the windows box itself_ it
wouldn't let you mount, even if perms were open to everyone.

As for the stale NFS handle, I have had similar problems with NFS in the
recent past. I spent ages trying to figure it out, and in the end it was
simply that I had to be someone other than root.

The other thing that helped was to make the exported file system use
no_root_squash.

Finally, you could always try the "Microsoft Solution" and reboot both
the Windows machine and the Solaris box. I've had that work before for
this kind of thing, too.

Sorry I can't be more directly helpful.

Paul.


On Thu, 2003-03-20 at 17:18, Michael French wrote:
>     Don't ask why, but I have to get a Windows box and a Solaris box talking
> over NFS.  The win box is the server and solaris box is the client.  I
> installed the Services for Unix on the Win box and configured NFS (what
> little there is to do).  I created a share and thought I was ready to rock.
> I do a showmount -e on the Win box and the share shows up.  rpcinfo
> servername shows all the processes I expect.
>     From the Solaris client, I can do the same thing with rpcinfo and
> showmount -e servername and get results back.  I had the firewall admin open
> up all IP traffic between these to boxes (just for now, for testing, having
> problems before this with just 2049 and 111 open).  I can mount the share
> with no problem (mount -o rw -F nfs 192.168.0.10:/testmount /testmount).  It
> mounts right away and running mount shows it with read/write/setuid on
> server, but as soon as I try to write to the share, I get a "Stale NFS"
> error.  I check the permissions on the mounted folder, changed to 777 before
> I mounted, owned by nobody/nobody (I am root right now).  On the win server,
> the log is showing a successfull mount, no errors.  The permissions on the
> directory on the server are wide open for all users, including connecting
> NFS clients.  Any idea what the hell could be going on?
> 
> I am using Solaris 8 and Win2K with SFU 2.2.  Win2K patched to SP3.
> 
> Thanks for any answers peope might be able to provide, I have to get this
> working today.
> 
> 
> Michael French
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-admin" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
-- 
Paul Furness

Systems Manager
Visual Information Lab
Mitsubsihi Electric ITE BV
Guildford, UK
 __________________________________________________________
|  Fight Spam! Join EuroCAUCE: http://www.euro.cauce.org/  |
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Re: Slightly off topic NFS question

[Michael French]

    Thanks for the tips Paul, unforntunately I had no luck in making this
work.  I am pretty sure it's an authentication issue, but I did have more
time to try and set the relationship with the MS SFU and the Unix client(s).
I dug around the net a little more and found a package called ProNFS which
works great.  I downloaded it, installed it, and had it running within 5
minutes.  Simple as can be, you select the directory you want to share,
assign users who can access it and set the access levels, all done on the
fly.  I was able to mount and copy files immediately.  Don't waste your time
with Services for Unix unless you have a week or two to play with it, the
instructions are horrible and the interface is completely non-intutive.

Michael French

----- Original Message -----
From: "Paul Furness" <paul.furness@vil.ite.mee.com>
To: "Michael French" <mfrench@ashevillemail.com>
Cc: <linux-admin@vger.kernel.org>
Sent: Thursday, March 20, 2003 11:10 AM
Subject: Re: Slightly off topic NFS question


> I'm afraid I can't give you a complete answer, just some pointers.
> Unfortunately, I haven't used NFS server from a windows machine, so
> these might be completely the wrong track. Anyway...
>
> I have had problems getting samba to mount windows shares in the past,
> and it seems likely that windows NFS uses some of the same auth
> mechanism. The main problem I had was that, if you tried to mount the
> windows share as a user that didn't exist _on the windows box itself_ it
> wouldn't let you mount, even if perms were open to everyone.
>
> As for the stale NFS handle, I have had similar problems with NFS in the
> recent past. I spent ages trying to figure it out, and in the end it was
> simply that I had to be someone other than root.
>
> The other thing that helped was to make the exported file system use
> no_root_squash.
>
> Finally, you could always try the "Microsoft Solution" and reboot both
> the Windows machine and the Solaris box. I've had that work before for
> this kind of thing, too.
>
> Sorry I can't be more directly helpful.
>
> Paul.
>
>
> On Thu, 2003-03-20 at 17:18, Michael French wrote:
> >     Don't ask why, but I have to get a Windows box and a Solaris box
talking
> > over NFS.  The win box is the server and solaris box is the client.  I
> > installed the Services for Unix on the Win box and configured NFS (what
> > little there is to do).  I created a share and thought I was ready to
rock.
> > I do a showmount -e on the Win box and the share shows up.  rpcinfo
> > servername shows all the processes I expect.
> >     From the Solaris client, I can do the same thing with rpcinfo and
> > showmount -e servername and get results back.  I had the firewall admin
open
> > up all IP traffic between these to boxes (just for now, for testing,
having
> > problems before this with just 2049 and 111 open).  I can mount the
share
> > with no problem (mount -o rw -F nfs 192.168.0.10:/testmount /testmount).
It
> > mounts right away and running mount shows it with read/write/setuid on
> > server, but as soon as I try to write to the share, I get a "Stale NFS"
> > error.  I check the permissions on the mounted folder, changed to 777
before
> > I mounted, owned by nobody/nobody (I am root right now).  On the win
server,
> > the log is showing a successfull mount, no errors.  The permissions on
the
> > directory on the server are wide open for all users, including
connecting
> > NFS clients.  Any idea what the hell could be going on?
> >
> > I am using Solaris 8 and Win2K with SFU 2.2.  Win2K patched to SP3.
> >
> > Thanks for any answers peope might be able to provide, I have to get
this
> > working today.
> >
> >
> > Michael French
> >
> > -
> > To unsubscribe from this list: send the line "unsubscribe linux-admin"
in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> --
> Paul Furness
>
> Systems Manager
> Visual Information Lab
> Mitsubsihi Electric ITE BV
> Guildford, UK
>  __________________________________________________________
> |  Fight Spam! Join EuroCAUCE: http://www.euro.cauce.org/  |
>  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>
>
>

gateway problem

[San]

hi!

i´m new on the list, and i don´t know if you have discussed this before or
if this is the place to post this kind of problems (if not, please correct
me) , but it s driving me insane, so here it goes...


i have a Red Hat 7.3 doing nat for an internal net of about 50 PCs.

it is doing so using shorewall (a wrapper for iptables), squid (virtual or
transparent mode) and a few things with iproute2.

I ´ve always had two internet connections, a direct connetion via a Cisco
Router(dont remember its model) and the other over ADSL.

I´d managed to have the two gateways running, using the direct connection
for inbound traffic and as a backup for outbound connections. All the
outbound trafic would go via ADSL, while it was connected and without
problems. If for a reason, the ADSL went down, a script would place the
direct connection as default gateway.

A brief description of IPs ant interfaces:

LAN
Gateway(3 NICs)
Internet

192.168.1.X
200.10.10.10             eth0      ---------------------------- Cisco
uter   --------------------------------------

200.10.10.11            eth0:0

192.168.1.1               eth1

N/A (or a fake one)
    -----------------------------ADSL          -----------------------------
------------

200.100.100.100       ppp0     ------------------|


The gateway has two external IPs (besides the ppp0 one). One is for
accessing the gateway itself, and the other for accessing an internal server
via DNAT.


All was working properly, but suddenly people who were using the internal
server from outside lost connection with it, even without the posibility of
ping'n it. Weird as it seemed to be working ok from inside.

Here is where im confused.

The ADSL is working ok, because it connects without problems, and NAT can do
its job serving internet for all the local network. But when it is
connected, PCs from outside cannot ping or access IPs binded to eth0 or
eth0:0(alias) and yes, they can ping and access services on ppp0.
If you disconnect the ADSL , and put back the gateway to have the direct
connection all start working ok again (but considering that the traffic on
the Router is higher than we want).

I checked a lot of things without success. The strange thing, leaving aside
the script that handles the changing default gateways (i mean, doing all by
hand), is that if the ADSL is connected but you keep the default gateway to
the Router, it works. The moment you change the DG, it's lost.

I'm sorry about the extension of this, but i couldn't find the way to
describe this using less words. I hope you could understand my english (i
know is bad) and if you need more details i would be more than happy to give
them to you...


i expect eagerly to hear from you all.

Thank You in advance!!!

Santiago Vazquez
Open Computacion S.A.
Buenos Aires
Argentina

-
To unsubscribe from this list: send the line "unsubscribe linux-admin" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html