Very Computer

Hi all,

Im having a problem using NFS. On my linux (RH7.2) server I export the /home
directory. I mount these directoires during startup on my linux clients in
/home using NFS. This seems to work fine. However, after some time the
client seems to hang when it tries to use the nfs mounted directories (on
the client). When I do (as root) a df command, the command seems to hang
when it is at the point of displaying the nfs mounted dir.
    Someone advised me to do this via autofs, but the problem remained the
same. The whole thing (client and server) uses NIS to do user auth.

What am I missing here?

TIA

Wilco

Quote:> Hi all,

> Im having a problem using NFS. On my linux (RH7.2) server I export the
/home
> directory. I mount these directoires during startup on my linux clients in
> /home using NFS. This seems to work fine. However, after some time the
> client seems to hang when it tries to use the nfs mounted directories (on
> the client). When I do (as root) a df command, the command seems to hang
> when it is at the point of displaying the nfs mounted dir.
>     Someone advised me to do this via autofs, but the problem remained the
> same. The whole thing (client and server) uses NIS to do user auth.

> What am I missing here?

> TIA

> Wilco

Are you turning off clients without unmounting things?

NFS is famous for problems in environments where the clients and servers pop
up and down: the "stateless" aspect of it makes things wait while the server
decides whether or not all the clients have done whatever they were doing
last with the mountpoints and files in them, and the timeouts are *long* and
unreliable.

Consider going to AFS if you can, and short of this be prepared to un-export
and re-export your mount points when this happens, and remount them on the
clients (basically done hby autofs if you need that, avoid am-utils!).

No, I take the clients down normally, that is: shutdown -h now (or poweroff)
It then tries to unmount everything including the nfs mounts....

Quote:

> NFS is famous for problems in environments where the clients and servers
pop
> up and down: the "stateless" aspect of it makes things wait while the
server
> decides whether or not all the clients have done whatever they were doing
> last with the mountpoints and files in them, and the timeouts are *long*
and
> unreliable.

> Consider going to AFS if you can, and short of this be prepared to
un-export
> and re-export your mount points when this happens, and remount them on the
> clients (basically done hby autofs if you need that, avoid am-utils!).

This is behaviour I'd expect so see with a server running IDE
disks and more than two or three clients mounting and working
on the disk at the same time...

What is your hardware?

Bj?rn
--
Bj?rn Tore Sund                  "When in fear, and when in doubt;
                                  Run in circles, scream and shout!"