Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.user > #162414 > unrolled thread

nfs problem

Back to article view | Back to linux.debian.user

Contents

  nfs problem ghe <ghe@slsware.net> - 2016-01-01 01:10 +0100
    Re: nfs problem Glenn English <ghe@srv.slsware.net> - 2016-01-01 23:00 +0100
      Re: nfs problem Pier <pierg75@yahoo.it> - 2016-01-02 00:50 +0100
        Re: nfs problem Glenn English <ghe@srv.slsware.net> - 2016-01-02 18:40 +0100
      Re: nfs problem Anders Andersson <pipatron@gmail.com> - 2016-01-02 11:00 +0100
        Re: nfs problem Glenn English <ghe@srv.slsware.net> - 2016-01-02 18:30 +0100
          Re: nfs problem [FIXED?] ghe <ghe@slsware.net> - 2016-01-03 00:10 +0100
    Re: nfs problem Brandon Vincent <Brandon.Vincent@asu.edu> - 2016-01-01 23:20 +0100
      Re: nfs problem Glenn English <ghe@srv.slsware.net> - 2016-01-02 18:40 +0100

#162414 — nfs problem

I'm trying to get NFS going with a Wheezy server and a Jessie client 
using Webmin and vi. It connects, but mounts the directory and file with 
nobody:nogroup as the user:group (and nobody isn't set to universe 
read/write -- tried to do that, but NFS won't let me).

I see from the 'Net that this is a common problem, but the solutions 
seem to involve (some) files/programs that don't exist on these 
machines. Nor in Aptitude.

The UID:GID are identical on the server and the client, both numbers and 
text. I've tried both NFS4 and vanilla NFS. There seems to be no difference.

I changed the nobody name to myself (ghe) in /etc/idmapd.conf, and 
created a new user (gheqw). Now ls says the directory on the client is 
owned by gheqw:nogroup. That doesn't make any sense to me at all.

When I turn off the NFS mount at the server, the directory on the client 
goes back to being owned by me, and the file disappears (both expected 
behavior).

There are too many variables here: 2 computers, many files, and far too 
many [SOLVED] pages on the 'Net. I'm pretty sure I've made a mess in the 
config files (whichever they are). Anybody know exactly what to say, in 
which file(s) on which computer(s), to deal with this?

T much IA...

-- 
Glenn English

[toc] | [next] | [standalone]

#162447

Come on folks!

Consider a modified Drake equation: (number on this list) * (date) * (% not hungover) * (% who understand NFS) * (% willing to help out a bewildered computer geek) == (surely > 0), no?

OK. Leave out (date) and (% not hungover), and tell me what I've done wrong...

> On Dec 31, 2015, at 4:56 PM, ghe <ghe@slsware.net> wrote:
> 
> I'm trying to get NFS going with a Wheezy server and a Jessie client using Webmin and vi. It connects, but mounts the directory and file with nobody:nogroup as the user:group (and nobody isn't set to universe read/write -- tried to do that, but NFS won't let me).
> 
> I see from the 'Net that this is a common problem, but the solutions seem to involve (some) files/programs that don't exist on these machines. Nor in Aptitude.
> 
> The UID:GID are identical on the server and the client, both numbers and text. I've tried both NFS4 and vanilla NFS. There seems to be no difference.
> 
> I changed the nobody name to myself (ghe) in /etc/idmapd.conf, and created a new user (gheqw). Now ls says the directory on the client is owned by gheqw:nogroup. That doesn't make any sense to me at all.
> 
> When I turn off the NFS mount at the server, the directory on the client goes back to being owned by me, and the file disappears (both expected behavior).
> 
> There are too many variables here: 2 computers, many files, and far too many [SOLVED] pages on the 'Net. I'm pretty sure I've made a mess in the config files (whichever they are). Anybody know exactly what to say, in which file(s) on which computer(s), to deal with this?
> 
> T much IA...
> 
> -- 
> Glenn English

-- 
Glenn English

[toc] | [prev] | [next] | [standalone]

#162451

[Multipart message — attachments visible in raw view] — view raw

On Friday, 1 January 2016, 21:57, Glenn English <ghe@srv.slsware.net> wrote:



 Come on folks!

Consider a modified Drake equation: (number on this list) * (date) * (% not hungover) * (% who understand NFS) * (% willing to help out a bewildered computer geek) == (surely > 0), no?

OK. Leave out (date) and (% not hungover), and tell me what I've done wrong...

> On Dec 31, 2015, at 4:56 PM, ghe <ghe@slsware.net> wrote:
> 
> I'm trying to get NFS going with a Wheezy server and a Jessie client using Webmin and vi. It connects, but mounts the directory and file with nobody:nogroup as the user:group (and nobody isn't set to universe read/write -- tried to do that, but NFS won't let me).
[big cut]
It was ages I didn't write to the list but nfs always arouse curiosity ;)First try to force nfsv3 to see if permissions are ok (v3 uses numeric uid/gid) with 'mount -o vers=3'.If this works and you want to use nfsv4 then make sure the configuration of idmap are the same on both the nfs server and nfs client (make sure the domain is the same, you can leave the rest the original).Also make sure the users/groups are available/the same on both the nfs server/client.
Hope this helps.
Pier
PS: you can check which version of nfs is used in /proc/mounts

   

[toc] | [prev] | [next] | [standalone]

#162488

> On Jan 1, 2016, at 4:41 PM, Pier <pierg75@yahoo.it> wrote:
> 
> It was ages I didn't write to the list but nfs always arouse curiosity ;)
> First try to force nfsv3 to see if permissions are ok (v3 uses numeric uid/gid) with 'mount -o vers=3'.
> If this works and you want to use nfsv4 then make sure the configuration of idmap are the same on both the nfs server and nfs client (make sure the domain is the same, you can leave the rest the original).
> Also make sure the users/groups are available/the same on both the nfs server/client.
> 
> Hope this helps.

So do I :-) I'm fine with v3. I'll try them.

-- 
Glenn English

[toc] | [prev] | [next] | [standalone]

#162459

On Fri, Jan 1, 2016 at 10:30 PM, Glenn English <ghe@srv.slsware.net> wrote:
> Come on folks!
>
> Consider a modified Drake equation: (number on this list) * (date) * (% not hungover) * (% who understand NFS) * (% willing to help out a bewildered computer geek) == (surely > 0), no?
>
> OK. Leave out (date) and (% not hungover), and tell me what I've done wrong...
>
>> On Dec 31, 2015, at 4:56 PM, ghe <ghe@slsware.net> wrote:
>>
>> I'm trying to get NFS going with a Wheezy server and a Jessie client using Webmin and vi. It connects, but mounts the directory and file with nobody:nogroup as the user:group (and nobody isn't set to universe read/write -- tried to do that, but NFS won't let me).
>>
>> I see from the 'Net that this is a common problem, but the solutions seem to involve (some) files/programs that don't exist on these machines. Nor in Aptitude.
>>
>> The UID:GID are identical on the server and the client, both numbers and text. I've tried both NFS4 and vanilla NFS. There seems to be no difference.
>>
>> I changed the nobody name to myself (ghe) in /etc/idmapd.conf, and created a new user (gheqw). Now ls says the directory on the client is owned by gheqw:nogroup. That doesn't make any sense to me at all.
>>
>> When I turn off the NFS mount at the server, the directory on the client goes back to being owned by me, and the file disappears (both expected behavior).
>>
>> There are too many variables here: 2 computers, many files, and far too many [SOLVED] pages on the 'Net. I'm pretty sure I've made a mess in the config files (whichever they are). Anybody know exactly what to say, in which file(s) on which computer(s), to deal with this?


Ok, I have a couple of suggestions, which should not necessarily be
implemented in this order. :)

1) Ditch webmin, I don't know what it is but it seems to break
something that should be pretty simple to set up, without giving any
feedback.
2) If the problem involves some files that do not exist, please list
them, maybe it's a clue that something important is missing.
3) NFS is well documented in the man-pages, but has a pathetic
documentation online. Try "man nfs" and "man exports".
4) As somone else mentioned, do not mess with idmapd, since you have
the UIDs synced already.
5) Until a few weeks ago my main NFS server was wheezy, but I am now
using stretch. My files looks like this:

On the (stretch) server: /etc/exports contains:
/srv/stuff   1.2.3.0/24(sec=sys,rw,no_subtree_check,mp,root_squash)

On the (jessie) clients: /etc/fstab contains:
name_of_server:/srv/stuff  /stuff  nfs4 _netdev,rw,auto,bg

You shouldn't have to change any other files, really, but I can't
remember if this has something that's not supported on wheezy.

[toc] | [prev] | [next] | [standalone]

#162485

> On Jan 2, 2016, at 2:56 AM, Anders Andersson <pipatron@gmail.com> wrote:
> 
> 1) Ditch webmin, I don't know what it is but it seems to break
> something that should be pretty simple to set up, without giving any
> feedback.

Webmin's one of those web-based GUI admin things. I use it because it usually does well, and it makes many configs quite a bit easier. It saved my life many times when I first hit Linux and IP networking.

In the past, Webmin's managed to make NFS work for me, but this seems to be one of those probs that require vi and a plea to the list...

> 2) If the problem involves some files that do not exist, please list
> them, maybe it's a clue that something important is missing.

I've looked at so many on the 'Net that I don't remember where I looked or what they were.

> 3) NFS is well documented in the man-pages, but has a pathetic
> documentation online. Try "man nfs" and "man exports".

Done, looks hopeful.

> 4) As somone else mentioned, do not mess with idmapd, since you have
> the UIDs synced already.

I'll put its config back to where it was :-)

> 5) Until a few weeks ago my main NFS server was wheezy, but I am now
> using stretch. My files looks like this:
> 
> On the (stretch) server: /etc/exports contains:
> /srv/stuff   1.2.3.0/24(sec=sys,rw,no_subtree_check,mp,root_squash)
> 
> On the (jessie) clients: /etc/fstab contains:
> name_of_server:/srv/stuff  /stuff  nfs4 _netdev,rw,auto,bg
> 
> You shouldn't have to change any other files, really, but I can't
> remember if this has something that's not supported on wheezy.

Very useful info -- different from what Webmin wrote. I'm a bit far from having an NFS export mounted at boot, though. When it works by hand, I'll put it in.

Thank you, Anders. 

-- 
Glenn English

[toc] | [prev] | [next] | [standalone]

#162511 — Re: nfs problem [FIXED?]

On 01/02/2016 10:06 AM, Glenn English wrote:

Beats me, but it's working.

I modified the line in /etc/exports (all on one line) to:

/home/ghe/Finances 
192.168.3.0/255.255.255.0(rw,no_root_squash,no_subtree_check)

That isn't exactly as suggested, but there were mild complaints about 
not having anything about subtree_check.

and in fstab on the client:

sbox.slsware.lan:/home/ghe/Finances
/home/ghe/Finances	nfs	user,nfsvers=3,exec,_netdev00

I'm thinking that 'nfsvers=3' might have done it, or maybe hand writing 
the export line containing the dotted quads for the IP/mask. Or maybe 
that I printed many pages of documentation. I really don't know; it just 
magically started working. I hate it when variables > 1.

Anyway, thank you all. I got my gnucash file in one place, on one 
server, but available to me everywhere on the LAN. And the next file 
will be a whole lot easier...

-- 
Glenn English

[toc] | [prev] | [next] | [standalone]

#162449

On Thu, Dec 31, 2015 at 4:56 PM, ghe <ghe@slsware.net> wrote:
> I changed the nobody name to myself (ghe) in /etc/idmapd.conf, and created a
> new user (gheqw). Now ls says the directory on the client is owned by
> gheqw:nogroup. That doesn't make any sense to me at all.

I would return the server's idmapd.conf back to the default
configuration (nobody:nogroup) and make sure that the idmapd service
is running on both the server and the client. Make sure the domain
matches in idmapd.conf on both the server and client.

Brandon Vincent

[toc] | [prev] | [next] | [standalone]

#162489

> On Jan 1, 2016, at 3:11 PM, Brandon Vincent <Brandon.Vincent@asu.edu> wrote:
> 
> I would return the server's idmapd.conf back to the default
> configuration (nobody:nogroup) and make sure that the idmapd service
> is running on both the server and the client. Make sure the domain
> matches in idmapd.conf on both the server and client.

Hmmm. Nobody told me before that I needed idmapd on both ends. Bears looking into. Thanks.

-- 
Glenn English

[toc] | [prev] | [standalone]

Back to top | Article view | linux.debian.user

csiph-web