White Paper

20
a Serviceguard NFS Modular Package” section of the Serviceguard NFS Toolkit
A.11.11.06, A.11.23.05 and A.11.31.08 Administrator’s Guide to configure the
/etc/rc.config.d/nfsconf file, start the NFS server, configure the disk hardware, and
setup volume groups, logical volumes, and file systems. It is assumed that you have
setup CFS as documented in the Veritas Storage Foundation Cluster File System
Installation and Administration Guide, and that you have already started the
Serviceguard CFS multi-node packages.
In the following example, the CFS file systems are /cfs1 and /cfs2, and they
correspond to Serviceguard CFS multi-node packages SG-CFS-MP-1 and SG-CFS-MP-2.
The cluster ASCII configuration file is assumed to be /etc/cmcluster/cluster.conf. The
cluster name is cluster1 and there are two nodes, thyme and basil. If you run the
cmviewcl command after setting up the Serviceguard CFS packages, the output is
similar to the following:
# cmviewcl
CLUSTER STATUS
cluster1 up
NODE STATUS STATE
thyme up running
NODE STATUS STATE
basil up running
MULTI_NODE_PACKAGES
PACKAGE STATUS STATE AUTO_RUN SYSTEM
SG-CFS-pkg up running enabled yes
SG-CFS-DG-1 up running enabled no
SG-CFS-MP-1 up running enabled no
SG-CFS-MP-2 up running enabled no
Create a directory on each server in the cluster to hold all the configuration files (if this
directory already exists you should save the contents before continuing):
# mkdir /etc/cmcluster/nfs_modular
The rest of the configuration is dependent upon whether or not the cluster requires file
locking (as described in the Issues and Limitations with the current CFS
implementation section). If file locking is not required, follow the instructions in section
“Serviceguard NFS Modular Package over CFS Packages without File Locking”. If file
locking is required, follow the instructions in section “Serviceguard NFS Modular
Package over CFS Packages with File Locking”.
7.1 Serviceguard NFS Modular Package over CFS Packages without File
Locking
Each active server in the cluster needs to run an export multi-node package and an
NFS failover package. An export multi-node package is a package that runs on each
server in the cluster and exports all the cluster file systems. Each standby server (i.e. a
server that is an adoptive node for NFS failover packages) needs to have an export
multi-node package running to be able to become active in the event of a failover.
Figure 5 shows an example Serviceguard over CFS configuration with two servers, each
with an NFS failover package and a multi-node export package.