One, straightforward, approach would be to configure a HA cluster of two nodes, which both have access to the shared storage (i.e. the lvols your NFS share(s) exist on). You would effectively have an NFS cluster package that has resources consisting of a virtual IP address and the NFS volume group. The package can run on either node, should one fail the package fails over to the other node and your NFS clients continue to access it through the VIP.
There are a number of different clustering solutions available, some only for certain distros. I guess it depends just how critical
this NFS data store is to you and whether you want vendor backed support. HP provide Serviceguard for Linux
- there is a license fee for the product and support (would need to check they agree to support your hardware), but they also have a free NFS toolkit (to do exactly what you want). Red Hat and SUSE also do their own clustering solutions and there are other community based options, such as DRBD
Hope this helps.