[gentoo-user] NFS setup

[gentoo-user] NFS setup

[Peter Humphrey]

On to the next problem...

This workstation serves as compute host to two smaller boxes on the network. I 
NFS-mount the PORTDIR of the smaller box in a chroot on this one, then do 
emerging and so on to build packages which I install later on the smaller box.

That works fine on one of the smaller boxes, but on the other I get a 
different port being used for NFS transfer every time. So I have to change the 
holes in the firewall before anything useful happens.

As far as I can see, everything to do with NFS is identical on the two boxes, 
in particular kernel config, /etc/conf.d/nfs and /etc/exports, but I must have 
missed something. SSH and SCP are not affected.

Any ideas?

-- 
Regards,
Peter.

Re: [gentoo-user] NFS setup

[Adam Carter]

[-- Attachment #1: Type: text/plain, Size: 559 bytes --]

On Sun, Aug 18, 2019 at 6:01 PM Peter Humphrey <peter@prh.myzen.co.uk>
wrote:

> On to the next problem...
>
> This workstation serves as compute host to two smaller boxes on the
> network. I
> NFS-mount the PORTDIR of the smaller box in a chroot on this one, then do
> emerging and so on to build packages which I install later on the smaller
> box.
>
> That works fine on one of the smaller boxes, but on the other I get a
> different port being used for NFS transfer every time.
>

 Is the output of 'mount | grep nfs' the same on the two client machines?

[-- Attachment #2: Type: text/html, Size: 896 bytes --]

Re: [gentoo-user] NFS setup

[Peter Humphrey]

On Sunday, 18 August 2019 09:30:36 BST Adam Carter wrote:

>  Is the output of 'mount | grep nfs' the same on the two client machines?

$ mount | grep nfs
nfsd on /proc/fs/nfsd type nfsd (rw,nosuid,nodev,noexec,relatime)

It's the same on both clients.

In the chroots, I see:

atom / # mount | grep nfs
192.168.1.2:/usr/portage on /usr/portage type nfs 
(rw,relatime,vers=3,rsize=131072,wsize=131072,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=sys,mountaddr=192.168.1.2,mountvers=3,mountport=32767,mountproto=udp,local_lock=none,addr=192.168.1.2)

(That's the one that works right)

clrn / # mount | grep nfs
192.168.1.4:/usr/portage on /usr/portage type nfs 
(rw,relatime,vers=3,rsize=1024,wsize=1024,namlen=255,hard,proto=tcp,timeo=600,retrans=2,sec=sys,mountaddr=192.168.1.4,mountvers=3,mountport=32767,mountproto=udp,local_lock=none,addr=192.168.1.4)

The only differences I can see are the IP addresses (of course) and the rsize 
and wsize; the good one has 128K, the other 1K. I'd better look into that, 
though it doesn't look like the problem. The good client is a 32-bit single-
core Atom, the other is a 64-bit quad-core Celeron.

/etc/conf.d/nfs is the same on both clients:
OPTS_RPC_NFSD="1"
OPTS_RPC_MOUNTD="-p 32767"
OPTS_RPC_STATD="-p 32765 -o 32766"
OPTS_RPC_IDMAPD=""
OPTS_RPC_GSSD=""
OPTS_RPC_SVCGSSD=""
OPTS_RPC_RQUOTAD=""
EXPORTFS_TIMEOUT=30

Any other ideas?

-- 
Regards,
Peter.

Re: [gentoo-user] NFS setup

[Adam Carter]

[-- Attachment #1: Type: text/plain, Size: 1190 bytes --]

On Sun, Aug 18, 2019 at 7:11 PM Peter Humphrey <peter@prh.myzen.co.uk>
wrote:

> On Sunday, 18 August 2019 09:30:36 BST Adam Carter wrote:
>
> >  Is the output of 'mount | grep nfs' the same on the two client machines?
>
> $ mount | grep nfs
> nfsd on /proc/fs/nfsd type nfsd (rw,nosuid,nodev,noexec,relatime)
>
>
nfs4 requires less ports than nfs3, just 2049 and something for mountd
(IIRC). Try using nfs4 and setting up the firewall for 2049 and 32767 from
your OPTS_RPC_MOUNTD="-p 32767" setting. From tcpdump, where .2 is the
client and .250 is the server;
192.168.1.2.949 > 192.168.1.250.2049: Flags [S]
but the other session is
192.168.1.250.730 > 192.168.1.2.40895: Flags [S]
ie a low port on the nfs server makes a connection back to the client, so
its quite unconventional

FYI, here's what one of mine looks like
$ mount | grep nfs
192.168.1.250:/export/public on /mnt/public type nfs4
(ro,noatime,vers=4.0,rsize=1048576,wsize=1048576,namlen=255,soft,proto=tcp,timeo=600,retrans=2,sec=sys,clientaddr=192.168.1.251,local_lock=none,addr=192.168.1.250,_netdev)

$ grep nfs /etc/fstab
192.168.1.250:/export/public    /mnt/public        nfs4
 ro,_netdev,vers=4.0,soft,noatime     0 0

[-- Attachment #2: Type: text/html, Size: 1743 bytes --]

Re: [gentoo-user] NFS setup

[Peter Humphrey]

On Sunday, 18 August 2019 10:35:36 BST Adam Carter wrote:
> On Sun, Aug 18, 2019 at 7:11 PM Peter Humphrey <peter@prh.myzen.co.uk>
> 
> wrote:
> > On Sunday, 18 August 2019 09:30:36 BST Adam Carter wrote:
> > >  Is the output of 'mount | grep nfs' the same on the two client
> > >  machines?
> > 
> > $ mount | grep nfs
> > nfsd on /proc/fs/nfsd type nfsd (rw,nosuid,nodev,noexec,relatime)
> 
> nfs4 requires less ports than nfs3, just 2049 and something for mountd
> (IIRC). Try using nfs4 and setting up the firewall for 2049 and 32767 from
> your OPTS_RPC_MOUNTD="-p 32767" setting. From tcpdump, where .2 is the
> client and .250 is the server;
> 192.168.1.2.949 > 192.168.1.250.2049: Flags [S]
> but the other session is
> 192.168.1.250.730 > 192.168.1.2.40895: Flags [S]
> ie a low port on the nfs server makes a connection back to the client, so
> its quite unconventional
> 
> FYI, here's what one of mine looks like
> $ mount | grep nfs
> 192.168.1.250:/export/public on /mnt/public type nfs4
> (ro,noatime,vers=4.0,rsize=1048576,wsize=1048576,namlen=255,soft,proto=tcp,t
> imeo=600,retrans=2,sec=sys,clientaddr=192.168.1.251,local_lock=none,addr=192
> .168.1.250,_netdev)
> 
> $ grep nfs /etc/fstab
> 192.168.1.250:/export/public    /mnt/public        nfs4
>  ro,_netdev,vers=4.0,soft,noatime     0 0

Thanks Adam. One thing I forgot to say is that emerge --sync and eix-update 
work fine with file transfers; it's only when a file is opened remotely that a 
new port is used.

But! All this is moot now, because today as I'm running my usual update, 
everything is working just fine! I haven't changed anything, honest.

Thanks again for your help; if it goes wrong again I'll follow your 
suggestion.

-- 
Regards,
Peter.

Re: [gentoo-user] NFS setup

[J. Roeleveld]

On 18 August 2019 10:01:18 CEST, Peter Humphrey <peter@prh.myzen.co.uk> wrote:
>On to the next problem...
>
>This workstation serves as compute host to two smaller boxes on the
>network. I 
>NFS-mount the PORTDIR of the smaller box in a chroot on this one, then
>do 
>emerging and so on to build packages which I install later on the
>smaller box.
>
>That works fine on one of the smaller boxes, but on the other I get a 
>different port being used for NFS transfer every time. So I have to
>change the 
>holes in the firewall before anything useful happens.
>
>As far as I can see, everything to do with NFS is identical on the two
>boxes, 
>in particular kernel config, /etc/conf.d/nfs and /etc/exports, but I
>must have 
>missed something. SSH and SCP are not affected.
>
>Any ideas?

Check the file /etc/conf.d/nfs in the nfs server.
There are settings documented there (using -p and -o options to fix the port numbers for the different parts.

I had the same issue in the past, which is why I remember.
I solved it by moving all NFS stuff onto a seperate VLAN which is only for storage.

--
Joost
-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

[gentoo-user] nfs setup

[John Dangler]

I'm trying to get nfs setup between my gentoo boxes (both local)
on the server, grep NFS /usr/src/linux/.config (on the server) returns:
CONFIG_NFS_FS=m
CONFIG_NFS_V3=y
CONFIG_NFS_V4 is not set
CONFIG_NFS_DIRECTIO is not set
CONFIG_NFSD=m
CONFIG_NFSD_V4 is not set
CONFIG_NFSD_TCP=y

Are these settings right to get nfs working ?

Thanks for the input.

John D
2.6.12-r9



-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] nfs setup

[Michael Crute]

[-- Attachment #1: Type: text/plain, Size: 995 bytes --]

On 8/31/05, John Dangler <jdangler@atlantic.net> wrote:
> 
> I'm trying to get nfs setup between my gentoo boxes (both local)
> on the server, grep NFS /usr/src/linux/.config (on the server) returns:
> CONFIG_NFS_FS=m
> CONFIG_NFS_V3=y
> CONFIG_NFS_V4 is not set
> CONFIG_NFS_DIRECTIO is not set
> CONFIG_NFSD=m
> CONFIG_NFSD_V4 is not set
> CONFIG_NFSD_TCP=y
> 
> Are these settings right to get nfs working ?
> 
> Thanks for the input.
> 
> 
Those look ok but you will have to modprobe NFS and NFSD into the kernel 
NFSD only on the server to make it work. You will also need to emerge 
nfs-utils. Also make sure to add the NFS modules to your modules autoload 
file in case the power goes out and your are forced to reboot.

-Mike

-- 
________________________________
Michael E. Crute
Software Developer
SoftGroup Development Corporation

Linux, because reboots are for installing hardware.
"In a world without walls and fences, who needs windows and gates?"

[-- Attachment #2: Type: text/html, Size: 1306 bytes --]

RE: [gentoo-user] nfs setup

[John Dangler]

Mike~
Thanks for the input.  I'm going to emerge nfs-utils on the server and
client now.  I can't wait to get off this win machine!

John 

-----Original Message-----
From: Michael Crute [mailto:mcrute@gmail.com] 
Sent: Wednesday, August 31, 2005 10:09 PM
To: gentoo-user@lists.gentoo.org
Subject: Re: [gentoo-user] nfs setup

On 8/31/05, John Dangler <jdangler@atlantic.net> wrote:
I'm trying to get nfs setup between my gentoo boxes (both local)
on the server, grep NFS /usr/src/linux/.config (on the server) returns:
CONFIG_NFS_FS=m
CONFIG_NFS_V3=y
CONFIG_NFS_V4 is not set
CONFIG_NFS_DIRECTIO is not set 
CONFIG_NFSD=m
CONFIG_NFSD_V4 is not set
CONFIG_NFSD_TCP=y

Are these settings right to get nfs working ?

Thanks for the input.

Those look ok but you will have to modprobe NFS and NFSD into the kernel
NFSD only on the server to make it work. You will also need to emerge
nfs-utils. Also make sure to add the NFS modules to your modules autoload
file in case the power goes out and your are forced to reboot.

-Mike

-- 
________________________________
Michael E. Crute
Software Developer
SoftGroup Development Corporation

Linux, because reboots are for installing hardware.
"In a world without walls and fences, who needs windows and gates?" 


-- 
gentoo-user@gentoo.org mailing list