in ,

Tweaking Network File System NFS server

Tweaking Network File System NFS server

NFS

NFS is the Network File System, which allows files located upon a remote system to be accessed as if they were local, with only a few caveats, (such as locking).

Although there are more complex and featureful remote filesystems available these days NFS has been in use for such a long time it is something you're bound to experience sooner or later.

NFS-Server Kernel Processes

When the nfs-kernel-server process starts a number of children are spawned to handle incoming requests.

If you edit the file /etc/default/nfs-kernel-server you can change the number of children from the default to a higher number, which will allow more connections to be handled:

# Number of servers to start up
RPCNFSDCOUNT=8

Once you make that change you can reload the server:

# /etc/init.d/nfs-kernel-server reload

Synchronous vs. Asynchronous Writing

There are two different ways of mounting NFS:

  • Synchronously
    • Write operations block until they've completed.
  • Asynchronously
    • Write operations return once they've been received.

The former method is the safest and most reliable, but the other method is faster. If you're prepared to accept the (minor) risk of write operations being lost then you should mount asynchronously.

To specify this just add the sync or async flags to your mount options:

# mount precious:/tmp /mnt -o sync
# mount precious:/tmp /mnt -o async

Network Buffer Sizes

When you have an filesystem mounted you can view the mount details to view the size of the network buffers used for reading and writing to the remote server.

Simply run:

root@shelob:~# grep "nfs " /proc/mounts
precious:/tmp /mnt nfs rw,relatime,vers=3,rsize=524288,wsize=524288,namlen=255,...

The two highlighted entres contain the read and write-buffer sizes, respectively. Here you see they're in the order of 500k each.

These buffer-sizes can be specified explicity when mounting the filesystem, if the defaults are low for your current-kernel:

# mount precious:/tmp  /mnt -o rsize=65536,wsize=65536

NOTE: The maximum value is limited depending on the kernels on each side.

Modifying MTU Size for NFS

MTU stands for Maximum Transmission Unit, and it is the largest amount of data that can be passed in one Ethernet frame.

Typically an MTU will be set to 1500, because raising this can cause problems wiht routers and similar devices. If you're on a modern LAN though you can increase the size to allow more data to be transmitted per-frame.

You can see your MTU size in the output of the “ip link show eth0” command, and set it via:

# ip link set dev eth0 mtu 5000

Having a large value will almost certainly cause you issues routing outside your network – but within a LAN you should be OK with the change.

 Author: Steve Kemp, of  tweaked.io

What do you think?

Leave a Reply

Your email address will not be published. Required fields are marked *

      Optimizing nginx reverse proxy

      Optimizing nginx reverse proxy

      SCP Command line tool

      SCP Command line tool