HA Mode

The role of the appliance can be changed by setting the HA Mode. If HA (First) Mode or HA (Second) Mode is selected as the HA Mode, a prompt will appear reminding to add a shared IP. Changing the HA Mode will require a reboot, so after the details are set, click the Reboot button provided. Once the LoadMaster has rebooted, the HA menu option will be available in the System Configuration section provided the role is not “Non HA Mode”. HA will NOT work if both machines are specified the same.

When logged into the HA cluster, use the shared IP address to view and set full functionality to the pair. If you log into the direct IP address of either one of the devices the menu options are quite different (see menus below). Logging into one of the LoadMaster directly is usually reserved for maintenance.

When a LoadMaster is in HA mode, the following screen appears when you select the HA Parameters menu option.

HA Status

At the top of the screen, next to the time, icons are shown to denote the real-time status of the LoadMaster units in the cluster. There will be an icon for each unit in the cluster. You can open the WUI for the first or second HA unit by clicking the relevant status icon.

The possible icons are:

Green (with ‘A’)		The unit is online and operational and the HA units are correctly paired. The A in the middle of the square indicates that this is the active unit.
Green (without ‘A’)		The unit is online and operational and the HA units are correctly paired. The absence of an ‘A’ in the middle of the square indicates that this is the standby unit.
Red/Yellow		The unit is not operational. It may be offline or misconfigured. The unit is not ready to take over. It may be offline or incorrectly paired.
Blue		When the unit reboots more than 3 times in 5 minutes it moves into a pacified state. In this state the machine is only accessible using the direct machine WUI (not the shared WUI), and, it is not participating in any HA activity, that is, no changes from the active unit are received and it will not take over if the active unit fails. To clear the pacified state, the unit must remain stable (without rebooting) for at least five minutes. After this stability period, a subsequent reboot clears the pacified state and returns the unit to normal operation.
Grey		The machine is in an indeterminate state and may require a reboot to return to operation. In some cases, this may mean both machines are active, that is, both are set to active, and something has gone seriously wrong. Call Progress Kemp Support for assistance with this issue, if rebooting does not solve it.
No HA icons		If the HA status squares are not appearing in the WUI, it probably means that HA is not enabled. Go to System Administration and select the HA option. Ensure the HA Mode is set to either First or Second.

In HA mode each LoadMaster will have its own IP address used only for diagnostic purposes directly on the unit. The HA pair have a shared IP address over which the WUI is used to configure and manage the pair as a single entity.

HA Mode

If using a single LoadMaster, select Non-HA Mode. When setting up HA mode, one LoadMaster must be set to HA (First) and the other HA (Second). If they are both set to the same option, HA will not operate.

HA Timeout

The time that the Active machine must be unavailable before a switchover occurs. With this option, the time it takes an HA cluster to detect a failure can be adjusted from 3 seconds to 15 seconds in 3 second increments. The default value is 9 seconds. A lower value will detect failures sooner, whereas a higher value gives better protection against a DOS attack.

HA Initial Wait Time

How long after the initial boot of a LoadMaster, before the machine decides that it should become active. If the partner machine is running, then this value is ignored. This value can be changed to mitigate the time taken for some intelligent switches to detect that the LoadMaster has started and to bring up the link.

HA Virtual ID

When using multiple HA LoadMaster clusters on the same network, this value uniquely identifies each cluster so that there are no potential unwanted interactions.

As of the 7.2.36 release, the LoadMaster selects a virtual ID based on the shared IP address of the first configured interface (the last 8 bits). It is selected and displayed once both the shared address and the partner address have been set. You can change the value to whatever you want (in the range 1 – 255) or you can keep it at the value it already selected. Please ensure the virtual ID is unique on each LoadMaster on the network.

Use Broadcast IP address

By default, the LoadMaster uses an IP multicast address when sending CARP packets. Enabling this option forces the use of the IP broadcast address instead.

Switch to Preferred Server

By default, neither partner in a HA cluster has priority. So that when a machine restarts after a switchover, the machine becomes the standby and stays in that state until forced to Active. Specifying a preferred host means that when this machine restarts, it will always try to become active and the partner will revert to standby mode. If a preferred server is specified, this will result in a double failover event because if the active unit fails, the standby unit will take over as active and when the preferred unit comes back up it will take over as active.

HA Update Interface

The interface used to synchronize the HA information within the HA cluster.

Hard Reboot on link Failure

In LoadMaster firmware version 7.2.53, a new option, Hard Reboot on link Failure, was introduced. When the Hard Reboot on link Failure check box is enabled, the LoadMaster configured in HA reboots if any configured interface loses connectivity with the network (that is, experiences a link failure). The reboot occurs regardless of the LoadMaster's HA status (Primary or Standby).

The Hard Reboot on link Failure check box is available in the System Configuration > HA Parameters screen when both of these are true:

• High Availability (HA) is configured

• The Switch to Preferred Server option is set to No Preferred Server.

  Note: The Hard Reboot on link Failure check box will be unavailable, if you select a preferred server from the Switch to Preferred Server drop-down list.

You cannot have a preferred server if Hard Reboot on link Failure is enabled - if you did, it could lead to circular swapping between the active and standby LoadMaster units.

Force Partner Update

Immediately forces the configuration from the active to standby unit without waiting for a normal update.

Inter HA L4 TCP Connection Updates

When using L4 services, enabling updates will allow L4 connections to be maintained across a HA switchover by sharing the connection table. This option is ignored for L7 services.

L4 Sync Threshold

The minimum number of incoming packets that a connection needs to receive before the connection will be synchronized.

Default value: 3 seconds

Valid values range from the 1 to the <SyncPeriod-1>. For example, if the L4 Sync Period is set to 50, the maximum L4 Sync Threshold would be 49.

L4 Sync Period

A connection gets synchronized every time the number of its incoming packets modulus sync_period equals the threshold.

Default value: 50 seconds

Valid values range from the <SyncThreshold+1> to 255.

L4 Sync Refresh Period

This is the allowed difference time (in seconds) in the reported connection timer that triggers a new sync message.

Default value: 0

Valid values range from 0 to 10.

Inter HA L7 Persistence Updates

When using L7 services, enabling this option will allow persistence information to be shared between the HA partners. If an HA failover occurs, the persistence information will not be lost. Enabling this option can have a significant performance impact.

HA Multicast Interface

The network interface used for multicast traffic which is used to synchronize Layer 4 and Layer 7 traffic when Inter-HA Updates are enabled.

Use Virtual MAC Addresses

Enabling this option forces the MAC address to switch between a HA pair during a switchover which is useful when gratuitous ARPs (used in communicating changes in HA IP addresses to switches) are not allowed.