Highly Available NFS supports the following configurations:
Simple failover from an active NFS
server node to an idle NFS server node.
Failover from one active NFS server node to another
active NFS server node, where the adoptive node supports more than
one NFS package after the failover.
A host configured as an adoptive node for more than
one NFS package. The host may also be prevented from adopting more
than one failed package at a time.
Cascading failover, where a package may have up
to three adoptive nodes.
Server-to-server cross mounting, where one server
may mount another server's file systems, and the mounts
are not interrupted when one server fails.
These configurations are illustrated in the following sections.
Simple Failover to an Idle NFS Server |
 |
Figure 1-1 “Simple
Failover to an Idle NFS Server” shows a simple
failover from an active NFS server node to an idle NFS server node.
Node_A
is the primary node for NFS server package Pkg_1.
When Node_A fails,
Node_B adopts
Pkg_1. This means
that Node_B locally
mounts the file systems associated with Pkg_1
and exports them. Both Node_A
and Node_B must
have access to the disks that hold the file systems for Pkg_1.
A Host Configured as Adoptive Node for Multiple Packages |
|
Figure 1-3 “A Host
Configured as Adoptive Node for Multiple Packages” shows a three-node
configuration where one node is the adoptive node for packages on
both of the other nodes. If either Node_A
or Node_C fails,
Node_B adopts
the NFS server package from that node.
When Node_A
fails, Node_B
becomes the server for Pkg_1.
If Node_C fails,
Node_B will become
the server for Pkg_2.
Alternatively, you can set the package control option in the NFS
control script to prevent Node_B
from adopting more than one package at a time. With the package
control option, Node_B
may adopt the package of the first node that fails, but if the second
node fails, Node_B
will not adopt its package. The package control option prevents
a node from becoming overloaded by adopting too many packages. If
an adoptive node becomes overloaded, it can fail.
Cascading Failover, with Three Adoptive Nodes |
|
A package may be configured with up to three adoptive nodes.
Figure 1-4 “Cascading
Failover, with Three Adoptive Nodes” shows this configuration.
If Node_A fails,
Pkg_1 is adopted
by Node_B. However,
if Node_B is
down, Pkg_1 is
adopted by Node_C,
and if Node_C
is down, Pkg_1
is adopted by Node_D.
The adoptive nodes are listed in the package configuration file,
/etc/cmcluster/nfs/nfs.conf,
in the order in which they will be tried. Note that all four nodes must have access
to the disks for the Pkg_1
file systems.
Server-to-Server Cross Mounting |
|
Two NFS server nodes may NFS-mount each other's file
systems and still act as adoptive nodes for each other's
NFS server packages. Figure 1-5 “Server-to-Server
Cross Mounting”
illustrates this configuration.
The advantage of server-to-server cross-mounting is that every
server has an identical view of the file systems. The disadvantage
is that, on the node where a file system is locally mounted, the
file system is accessed through an NFS mount, which has poorer performance
than a local mount.
Each node NFS-mounts the file systems for both packages. If
Node_A fails,
Node_B mounts
the filesystem for Pkg_1,
and the NFS mounts are not interrupted.
Printable version
