Regarding: Stale NFS File Handles on HPUX 10.20

Regarding: Stale NFS File Handles on HPUX 10.20

Post by Baby Pean » Sat, 27 Oct 2001 04:34:42



Ahh put they do for 10.20  It's a superceeded patch under
/superseded_patches/hp-ux_patches/s700_800/10.X/PHCO_9543 now
but it only has the umount static binary in it and no dependencies.

Patch Name:  PHCO_9543

Patch Description: s700_800 10.20 umount(1M) to unmount a Stale NFS filesystem

Creation Date: 96/12/12

Post Date:  96/12/17

Hardware Platforms - OS Releases:
        s700: 10.20
        s800: 10.20

Products: N/A

Filesets:
        OS-Core.UX-CORE

Automatic Reboot?: No

Status: General Superseded

Critical: No

Path Name:  /hp-ux_patches/s700_800/10.X/PHCO_9543

Symptoms:
        PHCO_9543:
        If a file system is mounted on system over NFS, and the
        file system goes down, doing a bdf will report and error
        of "Stale NFS File handler".  The current NFS umount
        will not allow this file system to be unmounted, even
        if the file system comes back on the original
        server.

Defect Description:
        PHCO_9543:
        The problem is caused by a bug in /sbin/fs/nfs/umount.  If
        a file system is mounted on over NFS, and the file system
        goes away for some reason, the file system will be reported
        as Stale, and there is no way to get rid of this without
        a reboot.

        To recreate the problem, make a directory on system A and
        export it.  On system B (must be 10.20), mount the file
        system.  Remove the directory on system A.  Do a bdf on
        system B, and it should report "Stale NFS file handler" for
        the mounted file system.  Doing a umount of the file system
        also reports this error, and the umount is not done.  Even
        if the directory is recreated on system A, and exported, it
        cannot be umounted.  The system would have to be rebooted.

SR:
        5003345868

Patch Files:
        /sbin/fs/nfs/umount

what(1) Output:
        /sbin/fs/nfs/umount:
                $Revision: 78.6 $
                PATCH/10.20:PHCO_8764 libc.a_ID@@/main/r10dav/libc_d
                        av/libc_dav_cpe/2
                /ux/libc/libs/libc/archive_pa1/libc.a_ID
                Oct  2 1996 10:35:25
                PATCH_10_20: /opt/langtools/lib/crt0.o umount.o 96/1
                        2/12

cksum(1) Output:
        2453477347 200704 /sbin/fs/nfs/umount

Patch Conflicts: None

Patch Dependencies:  None

Hardware Dependencies:  None

Other Dependencies:  None

Supersedes:  None

Equivalent Patches: None

Patch Package Size:  250 Kbytes

Installation Instructions:
        Please review all instructions and the Hewlett-Packard
        SupportLine User Guide or your Hewlett-Packard support terms
        and conditions for precautions, scope of license,
        restrictions, and, limitation of liability and warranties,
        before installing this patch.
        ------------------------------------------------------------
        1. Back up your system before installing a patch.

        2. Login as root.

        3. Copy the patch to the /tmp directory.

        4. Move to the /tmp directory and unshar the patch:

                cd /tmp
                sh PHCO_9543

        5a. For a standalone system, run swinstall to install the
            patch:

                swinstall -x autoreboot=true -x match_target=true \
                        -s /tmp/PHCO_9543.depot

        5b. For a homogeneous NFS Diskless cluster run swcluster on the
            server to install the patch on the server and the clients:

                swcluster -i -b

            This will invoke swcluster in the interactive mode and
            force all clients to be shut down.

            WARNING: All cluster clients must be shut down prior to the
                     patch installation.  Installing the patch while the
                     clients are booted is unsupported and can lead to
                     serious problems.

            The swcluster command will invoke an swinstall session in which
            you must specify:

                alternate root path  -  default is /export/shared_root/OS_700
                source depot path    -  /tmp/PHCO_9543.depot

            To complete the installation, select the patch by choosing
            "Actions -> Match What Target Has" and then "Actions -> Install"
            from the Menubar.

        5c. For a heterogeneous NFS Diskless cluster:

                - run swinstall on the server as in step 5a to install
                  the patch on the cluster server.

                - run swcluster on the server as in step 5b to install
                  the patch on the cluster clients.

        By default swinstall will archive the original software in
        /var/adm/sw/patch/PHCO_9543.  If you do not wish to retain a
        copy of the original software, you can create an empty file
        named /var/adm/sw/patch/PATCH_NOSAVE.

        Warning: If this file exists when a patch is installed, the
                 patch cannot be deinstalled.  Please be careful
                 when using this feature.

        It is recommended that you move the PHCO_9543.text file to
        /var/adm/sw/patch for future reference.

        To put this patch on a magnetic tape and install from the
        tape drive, use the command:

                dd if=/tmp/PHCO_9543.depot of=/dev/rmt/0m bs=2k

Special Installation Instructions:  None

> From: MFrk (molie...@hotmail.com)
> Subject: Re: Stale NFS File Handles on HPUX 10.20
> Newsgroups: comp.sys.hp.hpux
> View this article only
> Date: 2001-04-05 15:02:28 PST

> You will be able to unmount the stale nfs file system only prior doing a
> bdf...after doing a bdf, your only recourse is to reboot the
> device to clear the stale nfs file handle. I don't think HP has a fix for
> that.

> -M
> Alan D Johnson wrote in message <3ACC877C.48318...@ulster.net>...
> >Just umount them from the local machine. If that doesnt work then remove
> >the entries from /etc/fstab

> >Kote wrote:

> >> A number of NFS-mounted file systems have now become unavailable on a
> >> server, as they have been removed on the remote host server-which is all OK
> >> :)

> >> However, on the local server, when I do a "bdf", I get the message :

> >> Stale NFS file handle

> >> How do I get rid of these ? Coming from a Sun background, I know that I
> >> could stop and restart the nfs.client and that this would usually solve it.
> >> However, this does not seem to work on HPUX .

> >> Any ideas, anyone ?

> >> Thanks,

> >> Raymond

 
 
 

1. Stale NFS File Handles on HPUX 10.20

A number of NFS-mounted file systems have now become unavailable on a
server, as they have been removed on the remote host server-which is all OK
:)

However, on the local server, when I do a "bdf", I get the message :

Stale NFS file handle

How do I get rid of these ? Coming from a Sun background, I know that I
could stop and restart the nfs.client and that this would usually solve it.
However, this does not seem to work on HPUX .

Any ideas, anyone ?

Thanks,

Raymond

2. Deleteing Flash files

3. stale NFS handle after upgrade to 10.20

4. dual boot problem

5. Stale NFS file handles

6. Investment Opportunity!

7. Stale NFS File Handle

8. 19.13 core dump

9. Clearing Stale NFS file handles

10. Stale NFS file handles

11. Stale NFS handle

12. Stale NFS handles

13. Stale NFS handle on X-station