problem with NFS timeout options

Hello

I have great problem understanding the NFS mount options, and I wish
someone could clarify some things to me. Here is my situation:

I have a set of NFS servers, some up and running, others down, but I
don't know which is which. What I need, is a way to abort any mounting
effort from those servers that are dead as soon as possible (after, say,
1 second or so). My problem is that I cannot tune the timeout option
appropriately to get the desired behaviour. What happens, is that the
mount command hangs for some minutes before it eventually gives me an
error message.

Here is the command I used:

mount dead_server:/dir /mnt/dir -o soft,timeo=10,intr,retrans=0,retry=1

Now, according to the man pages, 'timeo' and 'retrans' settings should
have forced mount to terminate after 1 second. Otherwise, 'retry' option
should have taken over and stopped mount after one minute. Nevertheless,
neither happened.

I wonder what I understand wrong. Any ideas? By the way, is there a way
to test whether an NFS server is down, so that I don't try to mount from
it in the first place?


Tasos Kotaras
Senior Telecom Engineer
Intracom S.A.
Greece

Tasos Kotaras wrote:

> Here is the command I used:
>
> mount dead_server:/dir /mnt/dir -o soft,timeo=10,intr,retrans=0,retry=1
>
> Now, according to the man pages, 'timeo' and 'retrans' settings should
> have forced mount to terminate after 1 second. Otherwise, 'retry' option
> should have taken over and stopped mount after one minute. Nevertheless,
> neither happened.

Really you want to use "hard", not soft. this will cause a process to hang
if it tries to access a file on unavailable nfs mount - normally this is
what you want.

> I wonder what I understand wrong. Any ideas? By the way, is there a way
> to test whether an NFS server is down, so that I don't try to mount from
> it in the first place?

"showmount --exports hostname"

or you can just use the automounter to avoid all this.