Hi,
First many thanks to all people and excuse me for some mistakes or
precisions in my question. It was my fisrt query to Sun-manager, and I
promise to be more precise the next time.
Thanks to:
Yves Hardy yves@suntech.abcomp.be
Bismark Espinoza bismark@alta.jpl.nasa.gov
Bruce Rossiter arossite@us.oracle.com
Jerry Weber gldw@lanl.gov
Ayrton Sargusingh asargusi@sofkin.ca
Larry Boyd lboyd@isd.csc.com
Paul Chang paulc@sybase.com
Paulo Licio de Geu paulo@dcc.unicamp.br
Mark Anderson manderso@mitre.org
Problem : Stale NFS file handle remains on client after reboot of file server
>>Yves Hardy yves@suntech.abcomp.be
kill/restart any processes that have open file handles
>>Ayrton Sargusingh asargusi@sofkin.ca
If an NFS server goes offline for a few minutes,it should not
interfere with mountd so a 'stale NFS' condition should not occur.
>>Jerry Weber gldw@lanl.gov
Use mount with the -O option to o over-mount the same
mount point and device. This will see you thru until the next time that
you reboot.
>>Bismark Espinoza bismark@alta.jpl.nasa.gov
A drastic and not recommended way is to edit /etc/mnntab and remove
that filesystem, then mount it again.
>>Larry Boyd lboyd@isd.csc.com
There is a piece of share ware called lsof. get it compile it then
I use it like:
lsof | grep usr66
It lists the open filehandles holding a mount poit busy. Then kill them
or better still:
lsof | grep usr66 | awk '{print $2}' | xargs kill -9
This kills the processes using the mount point.
Conclusion:
In my case it's impossible to kill processes, so the best
solution is the over-mounting by Jerry Weber.
Effectively, a simple umounting and re-mounting a partition does not cause
a problem as explained by Ayrton Sargusingh. It occured some times during
a reboot or a crash.
Ps. As mentionned by some of yours , there was a mistake in my vfstab file.
It's was due to a (too rapid) change of an internal disk. Thanks
Ps2. As mentionned also,now I'm going to find the problem in my "From" field,
which has appeared on my Solaris machines only.
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+ Rene OCCELLI +
+ I.U.S.T.I. C.N.R.S. U.M.R. 139 +
+ Av. Esc. Normandie Niemen +
+ 13397 MARSEILLE Cedex 20 France +
+ Tel: (33)91 28 82 08 +
+ Fax: (33)91 28 82 25 +
+ Email: rene@iusti.univ-mrs.fr +
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Below my original mail
> Hi,
>
> Here a (simple ?) problem of mounted partitions which occurs on Solaris,
> but that i've never encoutered which Sunos.
>
> I've a server (Lisa) in Solaris 2.4
> Several partions are shared (/usr66 for exemple )
> Below a part of /etc/vfstab
> /dev/dsk/c0t1d0s6 /dev/rdsk/c0t3d0s5 /opt ufs 2 yes -
> and /etc/dfs/dfstab
> share -F nfs -o rw /usr66
>
> On other stations this partition are mounted.
> Exemple on Lili (solaris2.4)
> /etc/vfstab
> Lisa:/usr66 - /usr66 nfs - yes bg,hard,rw,retry=1000
>
>
> If for one reason the server crashes or simply reboot. This partition becomes
> iddle, even when the server (Lisa ) is Ok.
> Lili# ls /usr66
> /usr66: Stale NFS file handle
> Lili% df /usr66
> df: (/usr66 ) not a block device, directory or mounted resource
>
> The only way I've found is to unmount and remount the partition.
> But if a user is logged and use it. it's impossible to umount it. So
> the problem remains.
> Lili# umount /usr66
> nfs umount: /usr66: is busy
>
> This problen had never occured with Sunos. When a server is iddle, the client
> diplays the classic message :
> NFS server ...... still trying
> and all becomes Ok when the server is Ok again.
>
> Solutions?
> Thanks.
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:10:25 CDT