thewillhuang / hashistack Goto Github PK

HashiStack on AWS

This repo is inspired by kelseyhightower/hashiconf-eu-2016 to create hashistack(Consul/Nomad/Vault) cluster over AWS , easy bootstrap compare with kelseyhightower's original one .

Example setup

1 VPC
3 server subnet over 3 aws zone
3 client subnet over 3 aws zone
1 autoscaling group cross 3 aws zone and 3 server instances in each zone , fixed size
1 autoscaling group to  nomad client/agents , user can defined ASG size for nomad

Above subnets are all public and only accessed from your desktop public ip , you need setup your ip in dev.tfvars.example

Bootstrap

clone this repo

cd terraform

Generate ddns lambda function

zip -rj lambda_function_ddns.zip ../functions/ddns/*

edit dev.tfvars.example , update my_ip to your public ip/sshkey and other variable if needed.

terraform plan -var-file=dev.tfvars.example
terraform apply -var-file=dev.tfvars.example

You should able to see something like below output when terraform provision finished .

Outputs:

CONSUL_UI_ADDR = http://server-elb-000000000.us-east-1.elb.amazonaws.com:8500/ui
NOMAD_ADDR = http://server-elb-000000000.us-east-1.elb.amazonaws.com:4646
VAULT_ADDR = http://server-elb-000000000.us-east-1.elb.amazonaws.com:8200

To Use

Access consul UI <CONSUL_UI_ADDR>

Use Vault

export VAULT_ADDR=<VAULT_ADDR>
vault init
vault unseal
vault auth <root-token>

Run two system jobs

export NOMAD_ADDR=<NOMAD_ADDR>

make sure nodes are all ready

nomad node-status

As there is an issue with memory.limit_in_bytes issue on debian8 , I add reboot in userdata , it will make node up become slow as it needs reboot for the first time . If you choose ubuntu or other host os , you can just remove fix from client.sh

cd jobs
nomad run consul.nomad
nomad run fabio.nomad

Licnese

MIT.