HA (High Availability) LiteSpeed Load Balancer Cluster with Heartbeat

This how-to tutorial is tailored for RHEL/CentOS 5 distributions. It can also but be adapted to any other Linux/Unix distro.

Please note this how-to has been tested on a CentOS/RHEL 5 system with a heartbeat 2.1.3-3.el5 package. In Heartbeat's post 2.1.4 version releases, the software project has been reorganized. Although this how-to should work on the latest packages, it has not been tested.

For this guide we will setup 2 physical load balancer servers in an Active-Standby configuration using a single ethernet interface (eth0) for the HA heartbeat and shared IP.

Make sure that all the listener config in your two LiteSpeed LoadBalancer servers, as well as in any other place in the config where a source IP is configurable, are set to “ANY”.

lslb1.mydomain.com 5.5.5.20 #host name (uname -n) box one, this is primary Active
lslb2.mydomain.com 5.5.5.21 #host name (uname -n) box two, this is the secondary Standby

Make sure you use the hostname (uname -n) on the boxes as they appear exactly.

lslb1 is deisgnated the primary Active so that when both servers are restarted, rebooted, or for whatever reason, light up at the exact same moment, only lslb1 will acquire the shared IP resouce.

Shared IP resource

5.5.5.10 #this is the IP that we need to share and do a HA on. At any single moment, only one box will have this IP.

#download/install hearbeat rpm
yum install heartbeat

#enable heartbeat service on startup
chkconfig heartbeat on

Make sure you have the lslb init script in /etc/init.d and runnable via the command line “service lslb start”.

If you do not have “lslb” init file in /etc/init.d then install it:

#run the script that's in your lslb install directory to re-install the lslb init script if it does not exists
/lslbinstall/admin/misc/rc-inst.sh

Please note that heartbeat will consider itself down if “/etc/init.d/lslb stop” or “/etc/init.d/lslb start” comands fail for whatever reason. It is recommend you modify the return codes in the start) and stop) section of the init files to always return good “ret=0” where applicable.

There are 3 Hearbeat files we need to edit/create.

- /etc/ha.d/authkeys

auth 3
3 md5 myheartbeat_password

Change “myheartbeat_password” to a password you want heartbeat to use for communication between the nodes

- /etc/ha.d/haresources

lslb1.mydomain.com 5.5.5.10 lslb

List the primary Active server's hostname here and the shared IP resource. The last part of “lslb” tells heartbeat to run the script /etc/init.d/lslb with param “start” on resource acquisition and “stop” on resource dropping.

- /etc/ha.d/ha.cf

#heartbeat config
autojoin none
mcast eth0 239.0.0.43 694 1 0
bcast eth0
warntime 5
deadtime 10
initdead 15
keepalive 1
node lslb1.mydomain.com #again, make sure you use the exact hostname for each node
node lslb2.mydomain.com #...

#start service
service heartbeat start;

#check messages 
tail -f /var/log/messages;

#use this to see the ip address assigned to each server. Remember that the shared IP 5.5.5.10 should only show up 
#on one server at any moment. Test this by bring in each server down, taking down the heartbeat, taking down the #network cables, etc
ip addr sh eth0