Novell LINUX ENTERPRISE 11 - HIGH AVAILABILITY, Manual, Page: 103 / 298

9.2 High Availability
To construct a highly available Linux Virtual Server cluster, you can use several built-
in features of the software. In general, there are service monitor daemons running on
the load balancer to check server health periodically. If there is no response for a service
access request or ICMP ECHO_REQUEST from a server within a specified time, the
service monitor will consider the server dead and remove it from the available server
list at the load balancer. Thus, no new requests will be sent to this dead server. When
the service monitor detects that the dead server has recovered and is working again, it
will add the server back to the available server list. Therefore, the load balancer can
automatically mask the failure of service daemons or servers.
Furthermore, administrators can also use system tools to add new servers to increase
the system throughput or remove servers for system maintenance, without bringing
down the whole system service.
To prevent the load balancer from becoming a single point of failure for the whole
system, you need to set up a backup (or several backups) of the load balancer. Two
heartbeat daemons run on the primary and the backup, respectively. They periodically
heartbeat the “I'm alive” message to each other through serial lines and/or network in-
terfaces. When the heartbeat daemon of the backup cannot hear the heartbeat message
from the primary within the specified time, it will take over the virtual IP address in
order to provide the load-balancing service.
When the failed load balancer recovers again, there are two possible results: it can be-
come the backup load balancer automatically, or the active load balancer releases the
VIP address so that the recovered one takes over the VIP address and becomes the pri-
mary load balancer again. The primary load balancer has the state of connections,
meaning it knows which server the connection is forwarded to. If the backup load bal-
ancer takes over without that connection information, the clients have to send their re-
quests again in order to access service. In order to make load balancer failover transparent
to client applications, there is a connection synchronization in IPVS: the primary IPVS
load balancer synchronizes connection information to the backup load balancers through
UDP multicast. When the backup load balancer takes over after the primary one fails,
the backup load balancer will have the state of most connections, so that almost all
connections can continue to access the service through the backup load balancer.
Load Balancing with Linux Virtual Server 93