1745790 Members
4389 Online
108722 Solutions
New Discussion юеВ

Re: NFS stale

 
HCL Ash
Frequent Advisor

NFS stale

if we reboot nfs source.

Will auto mounter client gets stale NFS?

8 REPLIES 8
SoorajCleris
Honored Contributor

Re: NFS stale

Hi,

I think if you are already in the directory, yes, you may get.

I didnt check.

Regards,
Sooraj
"UNIX is basically a simple operating system, but you have to be a genius to understand the simplicity" - Dennis Ritchie
HCL Ash
Frequent Advisor

Re: NFS stale

it does not affect the auto mounter clients


Thx
Aneesh Mohan
Honored Contributor

Re: NFS stale


Hi,

Most propabaly you may get it ..

Aneesh
Dave Olker
HPE Pro

Re: NFS stale

A stale NFS directory or filesystem implies that the server has either un-shared the filesystem, the filesystem no longer exists on the server, or the filehandle used to represent this filesystem has somehow changed on the server.

If an NFS client has a filesystem mounted from a server and the server is rebooted, any access to that filesystem on the client will hang until the server begins responding to requests again. The client should not report a stale filehandle unless the server removes the shared directory, stops sharing the directory, or somehow modifies the filehandle used to represent the directory.

In a normal case, the server will reboot and the client will automatically recover with no user intervention.

Regards,

Dave
I work for HPE

[Any personal opinions expressed are mine, and not official statements on behalf of Hewlett Packard Enterprise]
Accept or Kudo
Viktor Balogh
Honored Contributor

Re: NFS stale

Last time I encountered some problems with automounter, the NFS server side was represented by a NetAPP vfiler. Where the shares have been actively used, the client HP-UX boxes needed to be rebooted to get freeed from the stale file handle
****
Unix operates with beer.
Dave Olker
HPE Pro

Re: NFS stale

A true stale filehandle should only occur for the reasons I stated earlier. I have no way of knowing why the NetApp filer was returning an ESTALE under normal working conditions, but it was likely one of those three scenarios.

If your HP-UX box is running 11i v2 or higher we support the forcible unmount option that should allow you to remove any hung/stale NFS filesystems without rebooting the client.

Regards,

Dave
I work for HPE

[Any personal opinions expressed are mine, and not official statements on behalf of Hewlett Packard Enterprise]
Accept or Kudo
Viktor Balogh
Honored Contributor

Re: NFS stale

>If your HP-UX box is running 11i v2 or higher we support the forcible unmount option

Dave: no, unfortunately we only have 11iv1 boxes. But I thought force-umount wasn't implemented until v3. At least it was stated so in the following white paper:

http://www.docs.hp.com/en/3929/ForciblyUnmountingNFSFilesystems.pdf

As for the NetAPP problem: a user had a cronjob running every minute in his home that was on a remote NFS share. So the mount was really busy, and after I umounted it a few seconds later it came back.
****
Unix operates with beer.
Dave Olker
HPE Pro

Re: NFS stale

Hi Viktor,

I guess that techncial paper is out of date. I blame the paper's author for being lazy.

Looking at the umount_nfs(1M) man page on 11i v2 systems you'll see:

-f Forcibly unmount a file system. Without this
option, umount does not allow a file system to be
unmounted if a file on the file system is busy.
Using this option can cause data loss for open
files; programs which access files after the file
system has been unmounted will get an error (
[EIO]).


At one time the techncial paper was correct, as 11i v2 didn't support forced unmount at initial release. However, we added this feature in PHKL_31500 Sept04 base patch:


( SR:8606352682 CR:JAGaf13487 )
( SR:8606353585 CR:JAGaf14379 )
Enhancement: This product update enables the forced unmount
of NFS file systems. When an NFS client loses contact with
a server, it is unable to recover without waiting for the
server to recover, at which point the client has a large
number of stale file handles. There is no option for
failing over to a working NFS server, since the client is
unable to unmount the NFS file system. The ability to do a
forced unmount is necessary.


Sorry for the confusion.

Dave

I work for HPE

[Any personal opinions expressed are mine, and not official statements on behalf of Hewlett Packard Enterprise]
Accept or Kudo