[nfs] mounting issues

Greetings !!

I have a local network using two computers: antares and sirius.
antares is the nfs-server and sirius is the nfs-client…

Precisions:

  • both firewalld are off on both computers (have no confidence, have always used iptables and ip6tables and my scripts running good on 42.3 can’t be used because “everything has changed” and did not had enough time to learn the new style ^^)
  • I always change the standard network ports (nfs is 4000 tcp), but keep smb and nmb ports.
  • rpc.mount is 4001

antares exported the different folders…

http://paste.opensuse.org/20415860

When using “mount -a” command on the client, for sure, the folders are mounted.
Same using KDE dolphin, each time I click on a umounted device in the list it is mounted.

BUT

When I boot my computer the devices are not mounted (???)
The smb share is not mounted
All the nfs exports are not mounted

http://paste.opensuse.org/41257606

http://paste.opensuse.org/3089083

The nfsmount.conf looks like this (client side – sirius)

http://paste.opensuse.org/97532814

…and like this (server side – antares)

http://paste.opensuse.org/39977661

The nfs.conf file seems to come from a template somewhere in the system but I guess it looks like this one (/etc/sysconfig/nfs):

http://paste.opensuse.org/69482368

Is there something I missed somewhere ?
As I read the online leap 15 documentation concerning NFS protocol and always used the same configuration for years (here the nfs is not tuned) without any dysfunctions…

(editing now because susepast didn’t worked as expected…)

I got “disallowed key characters” on sirius but it is working ok on antares with “same” configuration concerning firefox…

Don’t know why, second check indicates I got this message on both computer, … going crazy :{

As you found out yourself, your pastes are not very useful.

I do not quite understand why you do not copy/paste text output here in a post (of course between CODE tags. but you are long enough here to know that).

So from the NFS server

cat /etc/exports

and from the client

grep nfs /etc/fstab

and others that you think might be important.

You are also talking about smb (SAMBA). SAMBA is not NFS, thus I see no relation. When you have a SAMBA problem, it might be better to start a thread with at least the word SAMBA it the title.

/nfsshares/ntfs1        *(crossmnt,rw,no_root_squash)
/nfsshares/ntfs2        *(crossmnt,rw,no_root_squash)
/nfsshares/Datas1       *(crossmnt,rw,no_root_squash)
/nfsshares/Datas2       *(crossmnt,rw,no_root_squash)
/nfsshares/Datas3       *(crossmnt,rw,no_root_squash)
/nfsshares/Linux        *(crossmnt,rw,no_root_squash)
/nfsshares/System       *(crossmnt,rw,no_root_squash)


//192.168.0.3/exercices      /datas1/nfsshares/exercices  cifs     user=****,password=********  0  0

192.168.0.3:/windows/Ntfs1  /datas1/nfsshares/Ntfs1 nfs defaults 0 0
192.168.0.3:/windows/Ntfs2  /datas1/nfsshares/Ntfs2 nfs defaults 0 0  
192.168.0.3:/windows/Datas1 /datas1/nfsshares/Datas1 nfs defaults 0 0
192.168.0.3:/windows/Datas2 /datas1/nfsshares/Datas2 nfs defaults 0 0
192.168.0.3:/windows/Datas3 /datas1/nfsshares/Datas3 nfs defaults 0 0
192.168.0.3:/windows/Linux  /datas1/nfsshares/Linux nfs defaults 0 0
192.168.0.3:/windows/System /datas1/nfsshares/System nfs defaults 0 0

It took me more time to copy/paste than use susepaste command on both computers…

as I never had issues mounting after the system has booted, I never included /etc/fstab as a potential source of dysfunction…

antares:/nfsshares # ls -lah
total 8.0K
drwxr-xr-x  2 root root 4.0K Jun  6  2015 .
drwxr-xr-x 39 root root 4.0K Jul 21 08:28 ..
lrwxrwxrwx  1 root root   15 May 17  2015 Datas1 -> /windows/Datas1
lrwxrwxrwx  1 root root   15 May 17  2015 Datas2 -> /windows/Datas2
lrwxrwxrwx  1 root root   15 May 17  2015 Datas3 -> /windows/Datas3
lrwxrwxrwx  1 root root   14 May 17  2015 Linux -> /windows/Linux
lrwxrwxrwx  1 root root   15 May 17  2015 System -> /windows/System
lrwxrwxrwx  1 root root   14 May 16  2015 ntfs1 -> /windows/Ntfs1
lrwxrwxrwx  1 root root   14 May 16  2015 ntfs2 -> /windows/Ntfs2

a) but that did not work at all
b) sorry that I bother you to make it difficult for you with the only goal to make it easier for your potential helpers.

And I guess it will take you about the same time to extend (next time you try to post computer command and ouput) your copy mouse sweep so that it includes one line more at the top and the bottom. Then it will include the prompt and command line and the prompt line after all output. Easier for all others. In thjs case e.g. it will confirm what is done on antares and what is done on sirius. (and you did that more or less in your last post, thus why not always?)

I do not know why you have this strange structure and I never did something like this even for testing. But my idea is that you need to have in the fstab the files that are in the exports. Same text.

Your NFS server will know nothing about /windows/Ntfs1, because that is not in it’s exports.

Also from some names as you use them, I get the strange feeling that on the server non-Linux file system are used for those exports. I have no idea what the surprises will be when doing that.

Sorry hcvv… it is hot here and I did not sleep well :{

antares:/ # more /etc/exports
/nfsshares/ntfs1        *(crossmnt,rw,no_root_squash)
/nfsshares/ntfs2        *(crossmnt,rw,no_root_squash)
/nfsshares/Datas1       *(crossmnt,rw,no_root_squash)
/nfsshares/Datas2       *(crossmnt,rw,no_root_squash)
/nfsshares/Datas3       *(crossmnt,rw,no_root_squash)
/nfsshares/Linux        *(crossmnt,rw,no_root_squash)
/nfsshares/System       *(crossmnt,rw,no_root_squash)

sirius:/etc # showmount -e antares.dezordi.world 
Export list for antares.dezordi.world:
/windows/System *
/windows/Linux  *
/windows/Datas3 *
/windows/Datas2 *
/windows/Datas1 *
/windows/Ntfs2  *
/windows/Ntfs1  *
sirius:/etc # more fstab

... (things are there but not really needed) ...

192.168.0.3:/windows/Ntfs1  /datas1/nfsshares/Ntfs1 nfs defaults 0 0
192.168.0.3:/windows/Ntfs2  /datas1/nfsshares/Ntfs2 nfs defaults 0 0  
192.168.0.3:/windows/Datas1 /datas1/nfsshares/Datas1 nfs defaults 0 0
192.168.0.3:/windows/Datas2 /datas1/nfsshares/Datas2 nfs defaults 0 0
192.168.0.3:/windows/Datas3 /datas1/nfsshares/Datas3 nfs defaults 0 0
192.168.0.3:/windows/Linux  /datas1/nfsshares/Linux nfs defaults 0 0
192.168.0.3:/windows/System /datas1/nfsshares/System nfs defaults 0 0

^ those lines where added using nfs-client from Yast2

Well, it seems that one way or another those symlinks ae indeed interpreted by the server.
Not that I now understand why you use them and not simply export /windows/…

You say this construct worked before 15.0?

You say you have no firewall on?

Nothing in the boot logs / dmesg?

Random thoughts:
The system gets its list of what and when to mount during boot from /etc/fstab, which is why it is the start place for diagosing mount problems.
Without the the command line included in your pastes I cannot easily ascertain where they came from – including which machine.
Is the NFS server running before the client is booted?
Is the client using wicked for nertwork configuration and management or Network Manager? If the latter is the network connection dependent on user login?

I observed the same failure in opening paste.opensuse.org with Firefox (Falkon opened it fine, but not your links). Connecting via susepaste.org worked.

The nfs server has worked in 42.3 without problem during at least one year, the client did in 13.2 during four years (the server was there since 2008 and get versions from 11.2 to 13.2 to 42.3 until today)

The nfs server is called antares 192.168.0.3 and the client that can not mount at boot is called sirius (192.168.0.4)
Both firewalld are OFF.

For sure antares is booted before sirius, as I said NFS has worked for ten years.

Sirius can “see” the exported files using the command “showmount -e 192.168.0.3”

sirius:~ # showmount -e antares.dezordi.world 
Export list for antares.dezordi.world:
/windows/System *
/windows/Linux  *
/windows/Datas3 *
/windows/Datas2 *
/windows/Datas1 *
/windows/Ntfs2  *
/windows/Ntfs1  *

That works only if UI manually mount the system with “mount -a” command on sirius for sure…

sirius:~ # df -h
192.168.0.3:/windows/Ntfs1   513G  347G  166G  68% /datas1/nfsshares/Ntfs1                                                                                                                
//192.168.0.3/exercices       14G  5.3G  8.5G  39% /datas1/nfsshares/exercices                                                                                                            
192.168.0.3:/windows/Ntfs2   250G   72G  179G  29% /datas1/nfsshares/Ntfs2                                                                                                                
192.168.0.3:/windows/Datas1  160G   38G  123G  24% /datas1/nfsshares/Datas1                                                                                                               
192.168.0.3:/windows/Datas2  306G  272G   35G  89% /datas1/nfsshares/Datas2                                                                                                               
192.168.0.3:/windows/Datas3  380G   61G  319G  16% /datas1/nfsshares/Datas3                                                                                                               
192.168.0.3:/windows/Linux   276G  207G   70G  75% /datas1/nfsshares/Linux                                                                                                                
192.168.0.3:/windows/System  4.3G  3.4G  850M  81% /datas1/nfsshares/System  

What I want to do is understand why the system won’t mount them at boot…


sirius:/more /etc/fstab

192.168.0.3:/windows/Ntfs1  /datas1/nfsshares/Ntfs1 nfs defaults 0 0
192.168.0.3:/windows/Ntfs2  /datas1/nfsshares/Ntfs2 nfs defaults 0 0  
192.168.0.3:/windows/Datas1 /datas1/nfsshares/Datas1 nfs defaults 0 0
192.168.0.3:/windows/Datas2 /datas1/nfsshares/Datas2 nfs defaults 0 0
192.168.0.3:/windows/Datas3 /datas1/nfsshares/Datas3 nfs defaults 0 0
192.168.0.3:/windows/Linux  /datas1/nfsshares/Linux nfs defaults 0 0
192.168.0.3:/windows/System /datas1/nfsshares/System nfs defaults 0 0

This /etc/fstab worked since four years the previous computer connected to the server worked with this config since 2008, this computer has been replaced by sirius
that worked with this configuration in /etc/fstab since 2014.

the boolog indicates nothing particulary interresting… it includes ANSI sequences that messed up the possibility to copy/paste it…

See 'systemctl status datas1-nfsshares-Datas2.mount' for details.

See 'systemctl status datas1-nfsshares-Datas3.mount' for details.

See 'systemctl status datas1-nfsshares-Linux.mount' for details.

And the details are “could not mount …” that is very verbose…


antares:~ # netstat -anp --tcp | grep 4000
tcp        0      0 0.0.0.0:4000            0.0.0.0:*               LISTEN      -                   
tcp6       0      0 :::4000                 :::*                    LISTEN      -  

This is the nfsmount.conf on both computer (they are the same) I deleted the commented lines

 NFSMount_Global_Options ]

Defaultvers=3
Nfsvers=3
Defaultproto=tcp
Proto=tcp
Retrans=2
Retry=2
Background=True
Hard=True
Rdirplus=True
Rsize=8k
Wsize=8k
Bsize=8k
Timeo=600

mountport=4001

mountproto=tcp

# Server Mountd Version
mounvers=3


# Server Port (I never use the standard port numbers)
Port=4000


Extract of journactl on sirius (of course)

Jul 29 13:25:04 sirius systemd[1]: datas1-nfsshares-Ntfs1.mount: Unit entered failed state.
Jul 29 13:25:04 sirius systemd[1]: Failed to mount /datas1/nfsshares/Ntfs1.
Jul 29 13:25:04 sirius systemd[1]: datas1-nfsshares-Ntfs1.mount: Mount process exited, code=exited status=32
Jul 29 13:25:04 sirius systemd[1]: datas1-nfsshares-Ntfs2.mount: Unit entered failed state.
Jul 29 13:25:04 sirius systemd[1]: Failed to mount /datas1/nfsshares/Ntfs2.
Jul 29 13:25:04 sirius systemd[1]: datas1-nfsshares-Ntfs2.mount: Mount process exited, code=exited status=32
Jul 29 13:25:04 sirius systemd[1]: datas1-nfsshares-System.mount: Unit entered failed state.
Jul 29 13:25:04 sirius systemd[1]: Failed to mount /datas1/nfsshares/System.
Jul 29 13:25:04 sirius systemd[1]: datas1-nfsshares-System.mount: Mount process exited, code=exited status=32
Jul 29 13:25:04 sirius systemd[1]: datas1-nfsshares-Datas1.mount: Unit entered failed state.
Jul 29 13:25:04 sirius systemd[1]: Failed to mount /datas1/nfsshares/Datas1.
Jul 29 13:25:04 sirius systemd[1]: datas1-nfsshares-Datas1.mount: Mount process exited, code=exited status=32
Jul 29 13:25:04 sirius systemd[1]: datas1-nfsshares-Linux.mount: Unit entered failed state.
Jul 29 13:25:04 sirius systemd[1]: Failed to mount /datas1/nfsshares/Linux.
Jul 29 13:25:04 sirius systemd[1]: datas1-nfsshares-Linux.mount: Mount process exited, code=exited status=32
Jul 29 13:25:04 sirius systemd[1]: datas1-nfsshares-Datas3.mount: Unit entered failed state.
Jul 29 13:25:04 sirius systemd[1]: Failed to mount /datas1/nfsshares/Datas3.
Jul 29 13:25:04 sirius systemd[1]: datas1-nfsshares-Datas3.mount: Mount process exited, code=exited status=32
Jul 29 13:25:04 sirius systemd[1]: datas1-nfsshares-Datas2.mount: Unit entered failed state.
Jul 29 13:25:04 sirius systemd[1]: Failed to mount /datas1/nfsshares/Datas2.
Jul 29 13:25:04 sirius systemd[1]: datas1-nfsshares-Datas2.mount: Mount process exited, code=exited status=32
Jul 29 13:25:04 sirius kernel: FS-Cache: Netfs 'nfs' registered for caching
Jul 29 13:25:04 sirius systemd[1]: datas1-nfsshares-exercices.mount: Unit entered failed state.
Jul 29 13:25:04 sirius systemd[1]: Failed to mount /datas1/nfsshares/exercices.
Jul 29 13:25:04 sirius systemd[1]: datas1-nfsshares-exercices.mount: Mount process exited, code=exited status=32

error 32, from mountd is called “mount failure” ^^

systemctl enable NetworkManager-wait-online.service has been done… not started (it seems)
I reboot the computer to check…

it didn’t worked…

sirius:/datas1/installz/games/zerok/Zero-K/Saves # systemd-analyze critical-chain 
The time after the unit is active or started is printed after the "@" character.
The time the unit takes to start is printed after the "+" character.

graphical.target @13.631s
└─multi-user.target @13.631s
  └─smb.service @13.413s +217ms
    └─nmb.service @3.246s +10.164s
      └─network.target @3.240s
        └─NetworkManager.service @2.720s +277ms
          └─dbus.service @2.673s
            └─basic.target @2.666s
              └─paths.target @2.665s
                └─sendmail-client.path @2.665s
                  └─sysinit.target @2.639s
                    └─systemd-update-utmp.service @2.630s +8ms
                      └─auditd.service @2.605s +24ms
                        └─systemd-tmpfiles-setup.service @2.589s +15ms
                          └─local-fs.target @2.587s
                            └─root.mount @942ms +1.644s
                              └─local-fs-pre.target @918ms
                                └─lvm2-monitor.service @229ms +688ms
                                  └─lvm2-lvmetad.service @258ms
                                    └─lvm2-lvmetad.socket @221ms
                                      └─-.mount
                                        └─system.slice
                                          └─-.slice

Is / on SDD or NVME? Is NFS mounting being attempted before network is up? If yes, maybe adding nofail as a mount option would work around the delay? Network coming up late seems common complaint in recent months, maybe because booting is faster for more people switching from rotating rust.

I am sorry if you have already answered this and I have missed it, but I just want to be sure that the client (Sirius) has a network connection to the NFS server before any attempt to mount a remote filesystem is made. That is why I asked about wicked and NetworkManager.

I have encountered a similar situation with sshfs. One solution is to use autofs to mount on demand.

root filesystem “/” is on SDD yes…

I understand all the questions and this is not a problem for me to be more precise each time quesitons are asked, I’m just a little tired so the patience is not at a good level ^^

An option in the copy of the nfsmount.conf has be read as “mountvers” instead of “mounvers” (where the ‘t’ went ???)

I think autofs is not installed, I have to check… yes :{ it was not installed, so I think it will work next boot, I’m back.