This application is a loadbalancer that routes HTTP on public API requests only to Nodes that are currently available. The initial list of nodes provided in the configuration file is used as a starting point for a cluster discovery.
Sample config file, where port is pointing to a peer API:
network-nodes = [
{
host = "35.235.86.151"
port = "9001"
},
{
host = "34.94.105.94"
port = "9001"
},
{
host = "35.236.93.144"
port = "9001"
},
{
host = "35.235.119.9"
port = "9001"
},
{
host = "34.94.142.154"
port = "9001"
}]At least one of these nodes should always be operational and will serve as an entrypoint for cluster discovery.
Project is written in Scala and can be build with sbt. Building definiton includes task that provides a Docker image, ready to use for AWS deployment.
Before deployment this app should be published as a Docker image in the registry. TODO: provide registry
Once image is published, terraform script can be used to create, update or delete whole stack, including all dependencies and configuration.
Production configuration lives for now in the terraform/templates/application.conf file. It’s beeing automatically uploaded to S3 when the stack is applied.
Current stack definition deploys app to Fargate (ECS) on AWS, with autoscaling between 1 and 6 instances.
Once you update a Docker image (without changing it’s version) Terraform will not realise that it has to recreate ECS Task. To update project use taint command from terraform:
~/elb-manager on master! ⌚ 17:29:00
$ terraform taint aws_ecs_task_definition.appThis way you will redeploy the app itself and AWS will recreate tasks with new Docker image.
Once you change configuration file use Terraform apply command in the project root directory.
~/elb-manager on master! ⌚ 17:30:00
$ terraform apply terraform