Red Hat Knowledgebase

In most situations the default NFS configuration should allow file access at acceptable speeds. However there may be situations that require tuning of the NFS client for specific circumstances.

An example command to mount the remote export from a NFS server sharing large media files could be:

mount -t nfs -o noacl,nocto, rsize=32768,wsize=32768 mediaserver:/exports /mnt/mediaserver/

Below are possible mount options and their advantages and disadvantages:

noacl

Using this mount option the client will not attemp to to implement advanced Access Control Lists (ACL). This will save possible ACL read and write attempts when attempting to get advanced ACL permissions from a file.

Red Hat Enterprise Linux 3 fully supports NFS ACLs from the server. Red Hat Enterprise Linux 4 update 2 and below does not implement ACLs from the NFS server.

nocto

This option should reduce the NFS clients attempting to check or re-validate attributes on the NFS server. This mount option should only be used if the files are static and are not expected to change.

rsize

The default read default buffer size is 4096. Increasing this value may increase the performance depending on the size of the data being read. Recommended values for this parameter are numbers within the power of 2 (4096, 8192, ...). Large values may not work with NFS version 2.

wsize

The default read default buffer size of 4096, increasing this value may increase the performance depending on the size of the data being read.

In Addition to the listed Tuning parameters, consider the following parameters as well:

noatime

Setting this value disables the NFS server from updating the inodes access time.  As most applications do not necessarily need this value, you can safely disable this updating.

nodiratime

Setting this value disables the NFS server from updating the directory access time.  This is the directory equivalent setting of 'noatime'.

noac

This disables client side caching of inode attributes.  This effectively disables the defaults of acregmin=60 and acregmax=60 - 60 second timeouts for caching file attributes and directory attributes.  This can, actually cause a performance slowdown, but in certain circumstances can help if multiple NFS clients are actively writing to the same share.

Please note that there is no one-size-fits-all approach to performance tuning.  Different servers handling different workloads need to be tuned differently.  Please test and monitor both your server and client systems as you test different tuning options.