Consider two Linux machines:
Active machine heart beat IP: 192.168.1.1
Passive machine heart beat IP: 192.168.1.2
Virtual/Floating IP 192.168.2.11
Install heartbeat software on active and passive machines
manish@active-mc:~$ aptitude install heartbeat
manish@passive-mc:~$ aptitude install heartbeat
Get node names of both machines
manish@active-mc:~$ uname -n
active-mc
manish@passive-mc:~$ uname -n
passive-mc
On Active host do following
Configure ha.cf file
root@active-mc:/etc/ha.d# cat /etc/ha.d/ha.cf
# path of HA log file
logfile /var/log/ha-log
auto_failback on
# heartbeat signals port
udpport 694
node active-mc
# heartbeat network interface on primary
ucast eth1 192.168.1.1
node passive-mc
# heartbeat network interface on primary
ucast eth1 192.168.1.2
Configure resources to be covered under HA in haresources file
root@active-mc:/etc/ha.d# cat /etc/ha.d/haresources
#perferrered_machine Virtual IP service1 service2...
active 192.168.2.11 syslog-ng
remember that heartbeat looks for the service startup scripts in following directories and expects “start” argument to the startup scripts
/etc/ha.d/resource.d
/etc/init.d
If you want to use customize arguments to the startup script then you can write in it following format
active 192.168.2.11 myservice::myarg1
Add authentication key in authkeys file
root@active-mc:/etc/ha.d# cat /etc/ha.d/authkeys
auth1
1 sha1 a617248d47bce143ee169d17ef6298d9
Use a random hard to guess key. You can use simple tools like md5sum to generate key
tail -1000 /var/log/messages | md5sum
a617248d47bce143ee169d17ef6298d9
Copy ha.cf, haresources, authkeys files to Passive machine in /etc/ha.d/ directory
Start heartbeat service on both machines
/etc/init.d/heartbeat start
monitor /var/log/ha-log file on both machines.
To verify heartbeat function check if virtual IP is assigned on a network interface of primary host. In case you see it assigned on both active and passive machines then your heartbeat is not functioning properly. To troubleshoot the problem analyse traffic on network interface on both nodes configured in ha.cf.
tcpdump -i eth1 -vv -n port 694
Also verify that the services configured in HA resource are running on primary machine.