Veritas Storage Foundation 5.1 SP1 Cluster File System Administrator"s Guide (5900-1738, April 2011)

Understanding how Clustered NFS works
This Clustered NFS feature allows the same file system mounted across multiple
nodes using CFS to be shared over NFS from any combination of those nodes
without any loss of functionality during failover. The failover of NFS lock servers
includes all the locks being released by the old node then reclaimed by clients
talking to the new node during the grace period.
Basic design
The basic design is to have VCS manage Virtual IP (VIP) resources that can failover
between nodes and to add extra code into the steps used to handle these resources
to properly handle the NFS level operations. All other involved resources are
active on all nodes participating. The lock data, which is saved into a shared area,
is managed and used in lock-step with the Virtual IP resources to ensure that all
locks are reclaimed properly by clients while preventing any inappropriate locks
from being taken at the wrong time. Such interfering locks are prevented through
a combination of stopping services and using new features of the lock handling
inside the VxFS kernel code.
To communicate with the new code in the kernel, the fsclustadm command has
been modified to add command line interfaces to the private ioctl calls.
Note: You must have at least one VIP configured for each CNFS server.
Internal Clustered NFS functionality
This section describes the internal functionality of the triggers and actions scripts
that are a part of the Clustered NFS solution.
preonline trigger
The preonline script copies the lock state files created by the node status monitor
(normally called statd or rpc.statd daemon) during IP failovers and node failures.
The preonline script does the following on IP failover or node failure:
■ Finds the IP and the node it was last online on.
■ Finds the node on which the IP is next going to go online on.
■ Calls /opt/VRTS/bin/fsclustadm frlpause_enable and
/opt/VRTS/bin/fsclustadm frlock_pause to ensure that file system does
not give out any new locks during the failover.
Clustered NFS
Understanding how Clustered NFS works
142