Using SAMBA Toolkit in a ServiceGuard for Linux (SG/LX) Cluster Version A.01.00, June 2003

Using SAMBA Toolkit in a ServiceGuard for Linux (SG/LX) Cluster Version A.01.00
SAMBA Package Configuration Overview
Chapter 1 7
configuration files on a shared storage, it is recommended that the user
chooses all SMB/CIFS file systems and configuration files on the same
shared storage system.
Note: when a SAMBA package is extended to one or more additional
nodes, the same SMB/CIFS file system and configuration file structure
must be used on all nodes, regardless of whether this is a local or a
shared configuration.
Below is the note for a SAMBA local/shared configuration:
• Each node has the same version of the SAMBA application.
• Each node has the same configuration directory where identical
copies of the configuration files for each instance are placed in
local/shared hard disk.
• Each node has the same SMB/CIFS file system directories to access
the same shared storage.
Multiple SAMBA Instances Configuration
The toolkit can make the SAMBA server as a multiple-instance
application, which means that for more than one instance of the SAMBA
can be running on a node at the same time. For example, if two nodes are
running an instance of SAMBA and one node fails, the SAMBA instance
on the failed node can be failed over to the healthy node and it can
continue to run its own instance as well.
By bringing the SAMBA instances in service on one or more particular
nodes, the user can distribute the workload as desired. In addition, by
adjusting the failover policy for each of the instances, the user can
configure the SAMBA instance to fail over to a particular node, or to fail
over to different nodes to distribute the workload when a failure occurs.
By bringing the Apache instances in service on one or more nodes, the
user can distribute the workload as she/he like. And, by adjusting the
failover policy for each of the instances, the user can configure the
SAMBA instance to fail over to a particular node only, or to fail over to
different nodes to distribute the workload when a failure occurs.
There may be circumstances when the user wants to operate SAMBA in
a single instance mode, particularly if only one of cluster nodes is used
primarily for running SAMBA and other nodes are used for running