In parallel of the cluster deployment / management with InfraKit, we need to be able to join existing nodes to a cluster. That would be useful when the node is not on a supported cloud provider or when there's a prerequisite to have to nodes provisioned and administered by an organization (the customer's IT, a 3rd party provisioning process, etc).
The following actions should be performed:
install or update Docker to latest stable version
configure the Docker engine (labels, limits, TLS)
start or restart Docker
initialize or join a Swarm
if the Swarm has been initialized on this node, install the AMP services (Amplifier or the Amplifier proxy)
To do that, a binary (compiled from Golang for Linux|Darwin|Win / 64bit) should be provided. It can be a link on the AMP Cloudfront CDN, and / or a command line generated by the AMP CLI (with the arguments needed for the binary: the Swarm join IP and the join token).
ndegory
commented
7 years ago
In parallel of the cluster deployment / management with InfraKit, we need to be able to join existing nodes to a cluster. That would be useful when the node is not on a supported cloud provider or when there's a prerequisite to have to nodes provisioned and administered by an organization (the customer's IT, a 3rd party provisioning process, etc).
The following actions should be performed:
To do that, a binary (compiled from Golang for Linux|Darwin|Win / 64bit) should be provided. It can be a link on the AMP Cloudfront CDN, and / or a command line generated by the AMP CLI (with the arguments needed for the binary: the Swarm join IP and the join token).