# Tutorial: Set up high availability for Amazon Amazon IoT Amazon IoT Greengrass V2 with Pacemaker
<a name="pacemaker-tutorial"></a>

This tutorial shows you how to set up Amazon Amazon IoT Amazon IoT Greengrass V2 in a high availability (HA) configuration using [Pacemaker](https://clusterlabs.org/projects/pacemaker/), a high-availability cluster resource manager. You configure multiple Amazon EC2 instances as cluster instances, use [DRBD](https://linbit.com/drbd/) for block-level data replication, and manage Amazon IoT Greengrass V2 and load balancer services as Pacemaker resources with automatic failover.

**Important**  
This tutorial uses Amazon Elastic Compute Cloud instances to demonstrate the setup. You can deploy the Amazon IoT Greengrass V2 and Pacemaker integration to achieve high availability on a cluster of any device type, as long as the devices can communicate with one another.

This tutorial includes the following setups:
+ **Active/Passive Amazon IoT Greengrass V2 service** – Run Amazon IoT Greengrass V2 as a systemd service managed by Pacemaker with DRBD-replicated storage. Only one instance runs Amazon IoT Greengrass V2 at a time, and Pacemaker handles failover to a standby instance.
+ **Active/Passive load balancer** – Run HAProxy as a Pacemaker-managed resource with its configuration stored on DRBD-replicated storage. Pacemaker fails over the load balancer to a standby instance if the primary goes down.
+ **Active/Active Amazon IoT Greengrass V2 component** – Monitor a Amazon IoT Greengrass V2 component across all instances using a custom OCF (Open Cluster Framework) resource agent. Pacemaker detects component failures and triggers recovery without full instance failover.

Each setup is standalone and mutually exclusive. Each setup assumes a fresh start from the prerequisites, with the single DRBD resource repurposed for each setup's needs. Setup 3 (Active/Active) does not use DRBD — skip the DRBD prerequisite steps and install Amazon IoT Greengrass V2 to a local path on each instance instead.

In Setups 1 and 2, you create a highly available cluster of Amazon IoT Greengrass V2 devices. The cluster contains a *primary instance*, which is the instance that is currently active and running the managed services (such as Amazon IoT Greengrass V2 or HAProxy), and one or more *standby instances*, which are idle and waiting to take over if the primary fails. Pacemaker automatically promotes one of the standby instances to primary during failover. In Setup 3 (Active/Active), all instances run the service simultaneously, and Pacemaker handles per-instance recovery rather than failover promotion.

**Topics**
+ [Prerequisites and cluster setup](pacemaker-tutorial-prerequisites.md)
+ [Active/passive Amazon IoT Greengrass V2 service](pacemaker-tutorial-setup1.md)
+ [Active/passive load balancer](pacemaker-tutorial-setup2.md)
+ [Active/active Amazon IoT Greengrass V2 component](pacemaker-tutorial-setup3.md)