2003-02-17 16:40:50

by Santiago Flores

[permalink] [raw]
Subject: RedHat/Mandrake NFS - not playing nice

Hello. I usually don't mail lists without joining. Maybe I won't after
this either. Here is the problem:

Export filesytem (/var/www/html) from RedHat 7.3 to Mandrake 9.0 (mount
point /tmp/test or /r/amanda/nitro-html) fails with:

[root@backup amanda]# mount 10.0.0.48:/var/www/html /r/amanda/nitro-html
-t nfs
mount: 10.0.0.48:/var/www/html failed, reason given by server:
Permission denied
[root@backup amanda]#

However, there are no problems mounting the exported dir (or any others)
on other Redhat machines or debian machines (only unices handy).

The RedHat 7.3 is stock everything (stock nfs-utils, stock kernel
(non-enterprise)).
The Mandrake 9.0 is stock distro with kernel 2.4.19-ac4 (custom) with
NFS support compiled in.

/etc/exports:
/var/www/html 10.0.0.0/255.255.255.0(rw,no_root_squash)
#/var/lib/mysql/OldMaid 10.0.0.10/255.255.255.0(rw,no_root_squash)

rpcinfo -p (RedHat 7.3):
[root@chance proc]# rpcinfo -p
program vers proto port
100000 2 tcp 111 portmapper
100000 2 udp 111 portmapper
100024 1 udp 32768 status
100024 1 tcp 32768 status
391002 2 tcp 32769 sgi_fam
100011 1 udp 740 rquotad
100011 2 udp 740 rquotad
100011 1 tcp 743 rquotad
100011 2 tcp 743 rquotad
100005 1 udp 32817 mountd
100005 1 tcp 32940 mountd
100005 2 udp 32817 mountd
100005 2 tcp 32940 mountd
100005 3 udp 32817 mountd
100005 3 tcp 32940 mountd
100003 2 udp 2049 nfs
100003 3 udp 2049 nfs
100021 1 udp 32818 nlockmgr
100021 3 udp 32818 nlockmgr
100021 4 udp 32818 nlockmgr
[root@chance proc]#

rpcinfo -p (Mandrake 9.0):
[root@backup amanda]# rpcinfo -p
program vers proto port
100000 2 tcp 111 portmapper
100000 2 udp 111 portmapper
100011 1 udp 710 rquotad
100011 2 udp 710 rquotad
100011 1 tcp 713 rquotad
100011 2 tcp 713 rquotad
100003 2 udp 2049 nfs
100021 1 udp 33816 nlockmgr
100021 3 udp 33816 nlockmgr
100005 1 udp 33817 mountd
100005 1 tcp 32805 mountd
100005 2 udp 33817 mountd
100005 2 tcp 32805 mountd
[root@backup amanda]#

I have read the howto and googled for mandrake-RH NFS interaction but
not found anything of use to this problem. I have disabled ipchains and
iptables on both boxes and verfified that nothing is present in the
/etc/hosts.allow | /etc/hosts.deny files on either machine.

Like indicated above, the RH export works fine so long as I don't
attempt to mount it on a Mandrake box.

TIA,

Santiago




-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs


2003-02-17 19:16:25

by Trond Myklebust

[permalink] [raw]
Subject: Re: RedHat/Mandrake NFS - not playing nice

>>>>> " " == santi <Santiago> writes:

> The Mandrake 9.0 is stock distro with kernel 2.4.19-ac4
> (custom) with NFS support compiled in.

You appear to have forgotten to compile NFSv3 into this kernel. Try to
specify the option '-onfsvers=2'.

You appear also not to be running rpc.statd on that machine. Unless
you also use the 'nolock' mount option, you will end up with problems
the first time some process tries to lock a file.

Cheers,
Trond


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

2003-02-17 22:04:18

by Santiago Flores

[permalink] [raw]
Subject: Re: RedHat/Mandrake NFS - not playing nice

Thanks for responding so quickly Trond. I'm replying inline.

Trond Myklebust wrote:
>>>>>>" " == santi <Santiago> writes:
>>>>>
>
> > The Mandrake 9.0 is stock distro with kernel 2.4.19-ac4
> > (custom) with NFS support compiled in.
>
> You appear to have forgotten to compile NFSv3 into this kernel. Try to
> specify the option '-onfsvers=2'.

I recompiled the kernel to support all versions of NFS:

lqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqk
x
x x < > Coda file system support (advanced network fs)
x x < > InterMezzo file system
support (experimental,
x x <*> NFS file system support
x x [*] Provide NFSv3 client
support x x
<*> NFS server support
x x [*] Provide NFSv3 server support
x x [*] Provide NFS
server over TCP support (EXPERIMENTAL) x x
<*> SMB file system support (to mount Windows shares
x x [*] Use a default NLS
x x
Default Remote NLS Option: "cp437"
x x < > NCP file system support (to mount NetWare
volumes) x x
mqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqj
x
tqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqu
x <Select> < Exit > < Help >
x
mqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqqj


> You appear also not to be running rpc.statd on that machine. Unless
> you also use the 'nolock' mount option, you will end up with problems
> the first time some process tries to lock a file.

I took care of this too. Thanks for pointing it out.

[root@backup linux]# rpcinfo -p
program vers proto port
100000 2 tcp 111 portmapper
100000 2 udp 111 portmapper
100011 1 udp 682 rquotad
100011 2 udp 682 rquotad
100011 1 tcp 685 rquotad
100011 2 tcp 685 rquotad
100003 2 udp 2049 nfs
100003 3 udp 2049 nfs
100003 2 tcp 2049 nfs
100003 3 tcp 2049 nfs
100021 1 udp 32770 nlockmgr
100021 3 udp 32770 nlockmgr
100021 4 udp 32770 nlockmgr
100021 1 tcp 32796 nlockmgr
100021 3 tcp 32796 nlockmgr
100021 4 tcp 32796 nlockmgr
100005 1 udp 32771 mountd
100005 1 tcp 32797 mountd
100005 2 udp 32771 mountd
100005 2 tcp 32797 mountd
100005 3 udp 32771 mountd
100005 3 tcp 32797 mountd
100024 1 udp 32772 status
100024 1 tcp 32798 status


However, I still am receiving the same error as before:
[root@backup linux]# mount 10.0.0.48:/var/www/html /r/amanda/nitro-html
-t nfs
mount: 10.0.0.48:/var/www/html failed, reason given by server:
Permission denied
[root@backup linux]#

Any more ideas or suggestions?

Thanks,

Santiago



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

2003-02-17 22:29:02

by Paul Heinlein

[permalink] [raw]
Subject: Re: RedHat/Mandrake NFS - not playing nice

On Mon, 17 Feb 2003, Santiago wrote:

> However, I still am receiving the same error as before:
> [root@backup linux]# mount 10.0.0.48:/var/www/html /r/amanda/nitro-html
> -t nfs
> mount: 10.0.0.48:/var/www/html failed, reason given by server:
> Permission denied

I hate to ask the obvious questions, but

* Is there anything in /etc/host.deny that would keep your client from
getting to the server?

* Is there any sort of ipchains/iptables packet filtering that would
keep your client from getting to the server?

* Do the logs on the server offer any clues as to why it doesn't like
your client's mount requests?

--Paul Heinlein <[email protected]>



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs

2003-02-18 14:29:20

by Santiago Flores

[permalink] [raw]
Subject: Re: RedHat/Mandrake NFS - not playing nice

In my first post (you might not have seen it) I indicated that I had
checked the hosts.deny/hosts.allow and turned off ipchains and iptables.
It is usually an obvious thing. The obvious thing I forgot to do was to
check the log files. Here is what appears to be the issue:
Feb 17 16:19:15 chance rpc.mountd: refused mount request from 10.0.0.10
(private-10.swlink.net) for
/var/www/html (/var/www/html): no DNS forward lookup
Feb 17 16:22:07 chance rpc.mountd: authenticated mount request from
10.0.0.57:958 for /var/www/html
(/var/www/html)
Feb 17 16:22:16 chance rpc.mountd: refused mount request from 10.0.0.10
(private-10.swlink.net) for
/var/www/html (/var/www/html): no DNS forward lookup

RedHat is trying to look up the host that it is exporting to? I don't
have this issue though when mounting on a debian box. There is nothing
in the /etc/host file. So this is still somewhat mysterious. I am going
to put an entry in /etc/hosts for the mandrake box and see what happens.

Thanks reminding me. It is usually the obvious things.

It worked. Any ideas as to why the nfs server would differentiate
between the mandrake clients and the non-mandrake clients (RH and Debian)?

Thanks,
Santiago



Paul Heinlein wrote:
> On Mon, 17 Feb 2003, Santiago wrote:
>
>
>>However, I still am receiving the same error as before:
>>[root@backup linux]# mount 10.0.0.48:/var/www/html /r/amanda/nitro-html
>>-t nfs
>>mount: 10.0.0.48:/var/www/html failed, reason given by server:
>>Permission denied
>
>
> I hate to ask the obvious questions, but
>
> * Is there anything in /etc/host.deny that would keep your client from
> getting to the server?
>
> * Is there any sort of ipchains/iptables packet filtering that would
> keep your client from getting to the server?
>
> * Do the logs on the server offer any clues as to why it doesn't like
> your client's mount requests?
>
> --Paul Heinlein <[email protected]>
>
>
>
> -------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> NFS maillist - [email protected]
> https://lists.sourceforge.net/lists/listinfo/nfs
>
>




-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
NFS maillist - [email protected]
https://lists.sourceforge.net/lists/listinfo/nfs