Re: Some clients can't see NFS after re-boot - Unix

This is a discussion on Re: Some clients can't see NFS after re-boot - Unix ; Peter Smithson wrote: > Hi, > > I apologize in advance for not finding FAQ's etc. My excuse is that I'm > a developer and not an administrator and I don't have time to read up on > this issue ...

+ Reply to Thread
Results 1 to 4 of 4

Thread: Re: Some clients can't see NFS after re-boot

  1. Re: Some clients can't see NFS after re-boot


    Peter Smithson wrote:
    > Hi,
    >
    > I apologize in advance for not finding FAQ's etc. My excuse is that

    I'm
    > a developer and not an administrator and I don't have time to read up

    on
    > this issue properly.
    >
    > Our sys admin chap re-booted an NFS server on Friday and then went on


    > holiday for 2 weeks and is not contactable. We have about 10 UNIX
    > clients running different types of UNIX and some of them now can't

    see
    > the mounted directories.


    I suspect that the mounts that aren't working are not set to automount.

    >
    > I tried re-booting one of them but that didn't work - that's almost

    the
    > end of my sysadmin knowledge!
    >
    > They can all ping the server and they can all see another nfs server.
    >
    > The machines that work are -
    >
    > 3 Solaris boxes (various revs)
    > 2 AIX boxes
    > 1 HPUX 11 PARISC box
    >
    > The ones that have a problem are
    > 2 Linux boxes (2.4 and 2.3 kernels)
    > 1 SCO Open Server Box
    > 1 HPUX 11 Itanium (although rather oddly, this one's having a problem


    > with only one mount point from the same server - other mount points

    are
    > fine)
    > 1 Unixware 7.11 box


    On the boxes that don't work - if you know the directories that aren't
    mounted, try the mount command manually:
    mount server:/remotmountpoint /localmountpoint

    >
    > There are a few others that I've not checked.
    >
    > They all use an automounter.
    >
    > The NFS server is a Red Hat Linux box running 2.4 kernel. There's
    > another similar NFS server that's working OK.
    >
    > I logged into the server and had a look at /etc/exports but it

    doesn't
    > mention all the clients anyway. Just one of them infact. The others


    > appear to be other nfs servers and two machines in another office - I


    > thought you just listed the machines there but I only have a very

    basic
    > understanding.


    If the manual mount works, you know that the mount point is exported
    properly. If not, than make sure the system is listed in the
    /etc/exports for that mount point.

    If the manual mount does not work, post the error message here.
    A likely cause could be that the NFS daemon is not running.

    >
    > I've not even used automount before so it might be something to do

    with
    > that for all I know - could it be a NIS problem if the automount list

    is
    > stored there? But our passwords are on NIS and they are working.
    >
    > Any help appreciated.


    Google might help for the with the topic of NFS HOWTO and the specific
    OS you are on.

    >
    > Cheers
    >
    > Peter
    >
    >
    >
    > --
    > http://www.beluga.freeserve.co.uk



  2. Re: Some clients can't see NFS after re-boot

    In article <1110818345.540635.274640@z14g2000cwz.googlegroups. com>,
    gbeeker@gmail.com says...

    > I suspect that the mounts that aren't working are not set to automount.


    Possible bot not likley - why would some be set up one way? This was
    working before the re-boot of the NFS server and has done for several
    years (depending on the client in question).

    > On the boxes that don't work - if you know the directories that aren't
    > mounted, try the mount command manually:
    > mount server:/remotmountpoint /localmountpoint


    This is where I get problems! I'm not sure what your favourite OS is
    but I'll try SCO OS for this example -

    That's odd!

    # mount sgnfs1:/mpPrerel /Prerel
    mount: / is already mounted, /Prerel is busy,
    or allowable number of mount points exceeded: Device busy (error
    16)

    But now I can see /Prerel rather than getting no files returned. It's
    worked for the other mount points too. Something is really odd.

    I went onto the HPUX 11 Itanium machine and got this -

    # /usr/sbin/mount sgnfs1:/mpPrerel /Prerel
    Permission denied
    # ls /Prerel
    #

    So that didn't work. This is the odd machine where some mounts are
    working, from "df" -

    /Macros (sgnfs1:/mpMacros ): 1171984 blocks 121987
    i-nodes

    That's another directory from the same NFS server. If I do "ls
    /Macros" I get a list of files back.

    I can't check out the other machines as I don't have root access to them
    until the guy with the password list comes back in tomorrow. (it's gone
    5.30pm here).

    >If the manual mount works, you know that the mount point is exported

    properly. If not, than make sure the system is listed in the
    /etc/exports for that mount point.

    I mentioned the exports file. It doesn't have hardly any machines in it
    - including all the ones that work and the SCO OS one that I've just got
    working due to a manual mount. (or the HPUX one that works for one
    mount point but not another). There must be some other way they are
    getting exported but I don't know how.

    Thanks for your help so far.

    Peter

    --
    http://www.beluga.freeserve.co.uk

  3. Re: Some clients can't see NFS after re-boot


    Peter Smithson wrote:
    > In article <1110818345.540635.274640@z14g2000cwz.googlegroups. com>,
    > gbeeker@gmail.com says...
    >
    > > I suspect that the mounts that aren't working are not set to

    automount.
    >
    > Possible bot not likley - why would some be set up one way? This was


    > working before the re-boot of the NFS server and has done for several


    > years (depending on the client in question).


    That sounds like something has changed with the disk setup on the NFS
    server -or- the local clients, and then the reboot broke the
    connection. This happens a lot in the sysadmin world. Sometimes we have
    to diagnose problems that are a result of someone else (admin or
    developer with 'root') making a change, and not telling the other
    admins.

    >
    > > On the boxes that don't work - if you know the directories that

    aren't
    > > mounted, try the mount command manually:
    > > mount server:/remotmountpoint /localmountpoint

    >
    > This is where I get problems! I'm not sure what your favourite OS is


    AIX is my favorite. I have some Solaris exp. too.
    > but I'll try SCO OS for this example -
    >
    > That's odd!
    >
    > # mount sgnfs1:/mpPrerel /Prerel
    > mount: / is already mounted, /Prerel is busy,
    > or allowable number of mount points exceeded: Device busy

    (error
    > 16)


    This is odd, mount thinks you are trying to mount /. I'd suggest 2
    things, first, umount everything that is remote mounted, and try manual
    mounting the directories you really need.
    Second, if you get this error again, see if you can increase the number
    of mount points allowed.

    Hopefully a SCO expert can suggest you post some configuration file to
    help you further.

    >
    > But now I can see /Prerel rather than getting no files returned. It's


    > worked for the other mount points too. Something is really odd.
    >
    > I went onto the HPUX 11 Itanium machine and got this -
    >
    > # /usr/sbin/mount sgnfs1:/mpPrerel /Prerel
    > Permission denied
    > # ls /Prerel


    Right here, are you root? Do a 'ls -ld /Prerel' and see the directory
    perms. You need to have write perms to mount.
    For example, as root I can mount this:

    root@aix7:/home/root>mount intpub1:/os /mnt
    root@aix7:df -k /mnt
    intpub1:/os 24084480 2916516 88% 31410 1% /mnt

    root@aix7:/home/root>umount /mnt
    root@aix7:/home/root>ls -ld /mnt
    drwxr-xr-x 2 bin bin 512 Sep 20 2002 /mnt

    root@aix7:/home/root>id
    uid=0(root) gid=0(system)
    groups=2(bin),3(sys),7(security),8(cron),10(audit) ,50(dba)
    Since I am a member of bin group, I can write (or mount) to /mnt.

    > #
    >
    > So that didn't work. This is the odd machine where some mounts are
    > working, from "df" -
    >
    > /Macros (sgnfs1:/mpMacros ): 1171984 blocks

    121987
    > i-nodes
    >
    > That's another directory from the same NFS server. If I do "ls
    > /Macros" I get a list of files back.
    >
    > I can't check out the other machines as I don't have root access to

    them
    > until the guy with the password list comes back in tomorrow. (it's

    gone
    > 5.30pm here).

    Once you have root on those machines, do a ls -ld /dir, to make sure
    the permissions of the mount point allow you to write. Sometimes a
    mount point is not created, and you simple need to create it before the
    mount can happen.
    >
    > >If the manual mount works, you know that the mount point is exported

    > properly. If not, than make sure the system is listed in the
    > /etc/exports for that mount point.
    >
    > I mentioned the exports file. It doesn't have hardly any machines in

    it
    > - including all the ones that work and the SCO OS one that I've just

    got
    > working due to a manual mount. (or the HPUX one that works for one
    > mount point but not another).


    This indicates a permission issue with the local system, post ls -ld
    /mountpoint output for the mointpoint(s) you are attempting.

    > There must be some other way they are
    > getting exported but I don't know how.


    >
    > Thanks for your help so far.
    >
    > Peter
    >
    > --
    > http://www.beluga.freeserve.co.uk



  4. Re: Some clients can't see NFS after re-boot

    In article <1110893589.200936.164180@o13g2000cwo.googlegroups. com>,
    gbeeker@gmail.com says...

    Thanks for all the tips - I've just got back from messing around with
    the admin guy from another office and I think we've found the problem.
    Even though we only have one sysadmin person in this office, he's
    managed to trip himself up. He recently changed all the domains around
    and he uses them in the exports file.

    We spotted in the exports file syntax like this -

    *.ux.blahblah.local(rw)

    That allowed all the machines in that domain access. But the name of
    that domain was changed recently. I'm not sure why anything worked
    given that the domain changed. I did a ping from the server to a client
    machine to get the new domain name and put that in. I ran "exportfs -
    a" to re-read the exports file and things seem to be OK now.

    The other nfs server still has the old domain in it's exports file but
    that seems ok - really odd but at least we can work until he's back from
    holiday.

    Cheers

    Peter

    --
    http://www.beluga.freeserve.co.uk

+ Reply to Thread