3-24 T-Series Traffic Director Installation, Configuration and Administration Guide
Chapter 3 Basic T1000/Unison Load-Balancing Configuration
Configuring Minimum Servers and Backup Path
When a configured percentage of services bound to a virtual server have failed, it is
desirable to have all subsequent client requests sent to a backup virtual server. If the
remaining services cannot handle the load, it is typically better to bring down the entire
virtual server and send all client requests to a backup virtual server.
The following procedures describe how to bypass the cluster in a Firewall Load Balancing
setup if less than two of three OSN blades are healthy.
Adjust the vserver listen policy to factor in the total cluster health threshold. The cluster
health threshold is calculated by subtracting one from the minimum servers, then rounding
up. For example in a 3-OSN cluster where at least two nodes should remain active at any
given time, the cluster health threshold percentage is 34% (1/3 = 33.3%)
> set lb vserver Ingress -listenpolicy "CLIENT.VLAN.ID.EQ(151) &&
(SYS.VSERVER(\"Ingress\").HEALTH.GT(34))"
Done
Configuring Failover Redundancy
1
The T1000 pair exposes a single Layer-3 virtual IP address to next hop routers, allowing for
redundancy and failover. One T1000 is designated the primary and the other is the
secondary.
The configuration in this section assumes the following:
• There are two T1000 load balancers for the Unison cluster.
• The basic management configuration has been performed on both T1000s (see
Chapter 2, “Accessing and Configuring the Management Plane.”
• All data plane cabling has been installed on both T1000s.
• The primary T1000 has been fully configured using the procedures in this document.
Use the add HA node and set HA node commands to configure the T1000 for High
Availability. These commands have the following syntax:
add HA node id IPAddress
set HA node [-haStatus haStatus] [-haSync {ENABLED | DISABLED}] [-haProp
{ENABLED | DISABLED}] [-helloInterval msecs] [-deadInterval secs]
[-failSafe {ON | OFF}]
The helloInterval is the interval in milliseconds at which the heartbeat messages are sent to
the peer node. The heartbeat messages are UDP packets sent to port 3003 of the other node
in an HA pair. Values for the hello interval must be in the range 200ms to 1000ms. The
default hello interval is 200ms. The deadInterval is the time interval in seconds after which
the peer node is marked DOWN if heartbeat packets are not received. The default dead
interval is 3s. Values must be in the range 3s to 60s.
For example:
>add HA node 1 10.8.130.32
Done
1. Failover redundancy requires up to 20Mbps of traffic under standard operations for synchronization of the load bal-
ancer persistence table.
To Next Page
Loading...
Need help?
General
