<!-- ""@(#)hanfs 1.2 97/10/03 Sun Microsystems Inc"" >
 
<title>Highly Available NFS</title>
<h1>Highly Available NFS</h1>

Sun  Cluster 2.0 bundles a  Highly Available NFS (HA-NFS) data service
with fault   monitoring  agents with  the base   framework. HA-NFS can
co-exist with Parallel   Database Services as   well as other  HA data
services within the same cluster in SC 2.0.  HA-NFS is plug compatible
with existing NFS on Solaris and  does not involve  any changes to the
NFS protocol or client side NFS.  Neither  does it require any changes
to NFS administration on network clients.<p>

The fault   monitoring agents provided with  HA-NFS  run two  types of
periodic probes against a peer server's NFS service.<p>

1. The probing   server does a   NULL RPC to   all of the peer  daemon
processes that are required to provide NFS service. These daemons are:
rpcbind, mountd, nfsd, statd, and lockd.<p>

2. The  probing server does  an end-to-end test. It  tries to mount an
NFS file system from the  peer server, and  then to  read and write  a
file in that file system. It does this end-to-end  test for every file
system that the peer server is currently sharing. Because the mount is
expensive, it is executed less often than the other probes.<p>

The   checks conducted  by  the NFS  fault  probes  to avoid  spurious
takeovers is quite sophisticated and take  into account a wide variety
of situations.<p>
