beautypg.com

Requirements, Failover operation – Apple Mac OS X Server (version 10.2.3 or later) User Manual

Page 607

background image

Tools for Advanced Administrators

607

Note: IP failover only allows a secondary server to acquire a primary server’s IP address. You
need additional software tools such as rsync to provide capabilities such as mirroring the
primary server’s data on the secondary server. See rsync’s man pages for more information.

Requirements

IP failover is not a complete solution, rather one tool you can use to increase your server’s
availability to your clients. In order to use IP failover you will need to set up the following
hardware and software.

Hardware

IP failover requires the following hardware setup:

m primary server

m secondary server

m public network (servers must be on same subnet)

m private network between the servers (additional network interface card)

See “Setting Up a Private TCP/IP Network” on page 561 for more information on private
networks.

Note: Because IP failover uses broadcast messages, both servers must have IP addresses on
the same subnet of the public network. In addition, both servers must have IP addresses on
the same subnet of the private network.

Software

IP failover requires the following software setup:

m unique IP addresses for each network interface (public and private)

m software to mirror primary server data to secondary server

m scripts to control failover behavior on secondary server (optional)

Failover Operation

When IP failover is active, the primary server periodically broadcasts a brief message
confirming normal operation on both the public and private networks. This message is
monitored by the secondary server.

m If the broadcast is interrupted on both public and private networks, the secondary server

initiates the failover process.

m If status messages are interrupted on only one network, the secondary server sends email

notification of a network anomaly, but does not acquire the primary server’s IP address.

Email notification is sent when the secondary server detects a failover condition, a network
anomaly, and when the IP address is relinquished back to the primary server.

LL0395.Book Page 607 Wednesday, November 20, 2002 11:44 AM