WildFire Cluster High Availability
Where Can I Use
This? | What Do I Need? |
High availability is a crucial advantage of WildFire appliance
clusters because HA prevents the loss of critical data and services.
An HA cluster copies and distributes critical data, such as analysis
results, reports, and signatures, across nodes so that a node failure
does not result in data loss. An HA cluster also provides redundant critical
services, such as analysis functionality, WildFire API, and signature
generation, so that a node failure doesn’t interrupt service. A
cluster must have at least two nodes to provide high availability
benefits. Cluster node failure doesn’t affect firewalls, because
firewalls registered to a failed node use the cluster registration
list to register with another cluster node.
Each of the two devices in the HA pair is configured by the user
as a primary and secondary appliance. Based on this initial priority
value configuration, WildFire also assigns an operational status
of active to the primary appliance and passive to the secondary
device. This status determines which WildFire appliance is used
as the point of contact for management and infrastructure controls.
The passive device is always synchronized with the active appliance
and is ready to assume that role should a system or network failure
occur. For example, when the primary appliance in an active state
(active-primary) suffers a failure, a failover event occurs and
transitions to a passive-primary state, while the secondary appliance
transitions to active-secondary. The originally assigned priority
value remains the same regardless of the status of the appliance.
Failover occurs when the HA pair is no longer able to communicate
with each other, becomes unresponsive, or suffers a fatal error.
While the WildFire HA pair will attempt to auto-resolve minor disruptions,
major events require user-intervention. Failover can also be triggered
when a controller is suspended or decommissioned by the user.
Do not configure a cluster with only one controller node.
Each cluster should have an HA controller pair. A cluster should
have a single controller node only in temporary situations, for
example, when you swap controller nodes or if a controller node
fails.
In a two-node cluster HA pair, if one controller node fails,
the other controller node cannot process samples. For the remaining
cluster node to process samples, you must configure it to function
as a standalone WildFire appliance: delete the HA and cluster configurations
on the remaining cluster node and reboot the node. The node comes
back up as a standalone WildFire appliance.
Three-node clusters operate a HA pair with the addition of server
node to provide additional redundancy. The server operates the same
database and server infrastructure services as a controller, but
does not generate signatures. This deployment enables the cluster
to function if a controller node fails.
Additional nodes that are added to a WildFire cluster function
as a worker or server node. The third node is automatically configured
as a server, while each subsequent addition is added as a worker.