SUMMARY: NFS write error 70 messages

From: jtcummi@sandia.gov
Date: Fri Jun 04 1993 - 04:06:29 CDT


Thanks to all those who replied. I should have read the FAQ.
The original question was;

> Should I worry about these messages?
>
> May 26 09:47:45 sass807 vmunix: NFS write error 70 on host ramoth fh 71a 1 a0000 4d804 7cc5ed34 a0000 2 30caf9f9
> May 26 14:01:15 sass807 vmunix: NFS write error 70 on host ramoth fh 71a 1 a0000 4d804 7cc5ed3b a0000 2 30caf9f9
>
> Where sass807 is a sparc 10/30 running sunos 4.1.3 and ramoth is
> my main server a sparc 2 running sunos 4.1.3.
>
> I ran showfh and found out what file was involved. Could the file be corrupted?
> Should I warn my users. In the past week I have seen this about 9 times on 5
> different files.

cgehman@mlb.semi.harris.com answered:

I had the same problem on one of my computers. I'm not sure what caused
it, but I suspect that it was hardware related to the server. I moved
the disk to other computer and the problem went away. The strange thing
to me was that on the server, there were no errors. I never found any
corrupt files, but it's quite possible/probable.

casper@fwi.uva.nl answered:

It is strange that showfh comes up with a file: error 70 says ESTALE:
stale NFS file handle. It happens when a file open on a client is
remove on the server or another client.

It is possible that showfh finds a new file with the same inum,
but with a different inode generation. It is possible that
showfh doens't check the generation, as that cannot be done without
reading the filesystem directly.

You can ignore this error, but some of your users might need
to be educated in the use of your systems. NFS only acts as
a standard Unix filesystem if all operations on a single file
are done on the same host.

kevin@uniq.com.au answered:

The poor system is trying to tell you that
a stale file handle is being used.

Either remount the file systems in question (to get a new handle) or start
using the automounter (preferred solution).
hkatz@NUCMED.MED.NYU.EDU answered:

If you are running your 10/30 dataless off the sparc 2 check if you have
the correct permissions in your /etc/exports file on the SS2.

odt@dcs.bellcore.com answered:

NFS write error 70 messages refer to a stale NFS file handle. In your
case, they are from the host ramoth. You should go to ramoth and find
out what is stale - usually a df will suffice - and correct that situation.

Once that is done, I think you'll find that these messages will go away.

Most others said:

Look in /usr/include/sys/errno.h and you will see the line

#define ESTALE 70 /* Stale NFS file handle */

A stale NFS handle using results when a client NFS mounts a file system
and the filehandle to the file is no longer correct. This can be caused by
a server being down for a prolonged period of time for example.

ESTALE 70 Stale NFS file handle
An NFS client referenced a file that it had opened but
that had since been deleted.

error 70 is a stale file handle. it means that some user is removing
the file/directory in which the file lives while another user (on
another machine or possible the same user in another window) still
has the file/directory open.

Thanks also to:

canuck@masc38.rice.edu
clout!balr!vpnet!trdlnk!mike@cs.uchicago.edu
ups!upstage!glenn@fourx.aus.sun.com
stern@sunne.east.sun.com
fabrice@pure.com

Jeff Cummings, Sandia National Laboratories, (505)-844-2094, jtcummi@sandia.gov



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:07:54 CDT