Nodes Rolling Restart
In this section, we will learn how to restart Avalanche nodes sequentially (a.k.a. perform a "rolling restart") to avoid Subnet downtime.
For this tutorial, we will use the local inventory provided by the Ansible Avalanche Getting Started repository. See Create a Local Test Network for details.
Always make sure you have the latest version of the collection installed. See Installation and upgrade.
Disable automatic restarts
By default, any change to the node/VMs versions or configurations will trigger a restart of AvalancheGo. This is fine for testing purposes but not recommended for Subnet validators, (especially for a small-size permissioned Subnet where each validator has a high weight).
To disable automatic restarts, set avalanchego_auto_restart to false in the avalanche_nodes.yml group variables file:
avalanchego_auto_restart: false
Trigger a rolling restart of the nodes
After an upgrade or configation change, use the rolling_restart playbook. This playbook will restart every node sequentially, waiting for the node to be bootstrapped on all chains and healthy on all tracked Subnets before moving to the next node.
ansible-playbook ash.avalanche.rolling_restart -i inventories/local
It can take a while for a node to be bootstrapped and healthy depending on the node hardware and the Subnet activity.