Robust NFS file handles

traene · 03-28-2007, 03:14 AM

Hi, when i try to access a file via NFS and the network interface shuts down,
while accessing this file, my process blocks.

After starting up the network again, my process seems to keep on reading where it stopped.

Are there any drawbacks? How much time can the network shutdown take until problems occure (ie. TCP connections with Keep Alive can be idle for about 2hrs and a few minutes)?

Are there some dependencies on mount flags? I use:

Code:

/etc/fstab:
HOST:/vol   /mnt/dir   nfs  defaults 0 0

indienick · 03-28-2007, 09:59 PM

I don't see any issue that may arise, nor have I heard of any.

The only place I could see this happening, is if you're on a really weak wireless or bluetooth network, or if the resident sysadmin doesn't like your network activity and decides to blockade you from LAN access. In wires I trust.

What sort of problems are you thinking that may happen if the connection gets terminated?

traene · 03-29-2007, 06:36 AM

The machines are in a live and productive environment. We want to do some maintenance work on the NFS server and need to shut it down. The server will not be available for a few minutes.

This can happen when the process on the slave host is accessing a file on the NFS server. So the question is actually:

Will the process always block if the NFS mount is not available?