Bootstrapping a Consul Cluster With Cloud-Init-Buddy

Bootstrapping a consul cluster is a non-trivial operation mostly because it requires sequencing the startup of the cluster nodes with the starting of the consul agents. You can't form a cluster until you know the IP addresses of the nodes to pass to the consul agent for registration and cluster formation. So you first need to start the nodes and then distribute the IP addresses to all the nodes that need to be in the cluster. This is where cloud-init-buddy comes in. Cloud-init-buddy provides the glue for the cluster nodes to discover each other and not worry about sequencing startup and setup. Everything happens in the cloud-init script with all the sequencing and discovery automatically handled by the cluster nodes themselves.

The first thing we need to do is create a cloud-init-buddy node. I'm going to use the one provided in the Digital Ocean VM, but you can use whatever cloud provider you feel most comfortable with. The only thing you need is for the VM to support cloud-init.

Here's the cloud-init configuration for setting up the cloud-init-buddy node

#!/bin/bash
while ! (git clone https://github.com/cloudbootup/cloud-init-buddy.git); do
  echo "Looks like network is not ready. Sleeping and retrying."
  sleep 2
done
cd cloud-init-buddy
sudo apt update && sudo apt install --yes ruby
rake setup:initialize
rake flyway:check || rake flyway:install
rake postgres:configure
npm install
./node_modules/.bin/tsc
node utils/generate-certificate.js
tmux new-session -d -s cloud-init-buddy 'node app.js' && sleep 1
node utils/users.js add-user admin \
  $(gpg2 --gen-random 0 64 | shasum -b -a 512 | head -c 32 | tee password)

Wait for the VM to come up. Once it is up, log in to find out the password for the admin user because we need the password for the consul nodes to talk to cloud-init-buddy to coordinate the cluster formation. You can get the admin password by running cat /cloud-init-buddy/password <(echo).

At this point, we have all the pieces and can write the cloud-init scripts for the consul cluster. I'm going to use the values that I got and you will need to substitute your own values for the IP address and admin password.

#!/bin/bash
# Initial parameters for interacting with cloud-init-buddy
export CLOUD_INIT_BUDDY="138.197.199.132"
export PASSWORD="4c6f44cd5b5d549731d9ec1e3d7179e9"
export ENDPOINT="https://admin:${PASSWORD}@${CLOUD_INIT_BUDDY}:8443"
export NODE_COUNT=3
# Directory for us to do work
mkdir /consul
cd /consul
while ! (wget https://releases.hashicorp.com/consul/1.0.6/consul_1.0.6_linux_amd64.zip); do
  echo "Waiting for network to be ready."
  sleep 2
done
apt install unzip
unzip -o *.zip
rm -f *.zip
# What is our IP address? We will need this for registration
export SELF_ADDRESS="$(ip route get 8.8.8.8 | awk '{print $NF; exit}')"
# Initialize the metadata for coordination
export CONSUL_ENDPOINT="${ENDPOINT}/metadata/consul"
curl -k -XPOST "${CONSUL_ENDPOINT}" -d '{"hosts":[]}'
# Wait for all nodes to register
while [[ "$(curl -k "${CONSUL_ENDPOINT}/hosts.length")" -lt "${NODE_COUNT}" ]]; do
  # It is possible some other node reset everything so make sure we re-register
  if ! (curl -k "${CONSUL_ENDPOINT}" | grep "${SELF_ADDRESS}"); then
    curl -k -XPOST "${CONSUL_ENDPOINT}/hosts" -d "\"${SELF_ADDRESS}\""
  fi
  echo "Waiting for other nodes to register."
  sleep 1
done
# Whoever registered first will set a key
if [[ "$(curl -k "${CONSUL_ENDPOINT}/hosts/0")" == "\"${SELF_ADDRESS}\"" ]]; then
  key="$(./consul keygen | head -c 24)"
  curl -k -XPOST -d "{\"key\":\"${key}\"}" "${CONSUL_ENDPOINT}"
fi
# Wait for the key to be set
while ! (curl -k "${CONSUL_ENDPOINT}.keys" | grep "key"); do
  echo "Waiting for key to be set"
  sleep 1
done
# Everyone is registered and there is a key so we can form the cluster.
# In a production setting this would be an actual systemd unit file 
# and you would not use public facing IP addresses, i.e. you'd run
# the nodes in a private address space
nohup ./consul agent -ui -syslog -server -bootstrap-expect "${NODE_COUNT}" \
  -data-dir "/consul" \
  -bind "${SELF_ADDRESS}" \
  -advertise "${SELF_ADDRESS}" \
  -encrypt "$(curl -k "${CONSUL_ENDPOINT}/key" | tr -d '"')" \
  -retry-join "$(curl -k "${CONSUL_ENDPOINT}/hosts/0" | tr -d '"')" &

It looks a little ugly but it gets the job done.

Cloud-init-buddy also provides an endpoint for storing and retrieving tar packages but I didn't use that feature here. It comes in handy when you want to include other assets on the consul nodes or if you want to use something other than bash for bootstrapping the node. The package could include a Ruby or Python script and the cloud-init script would just be reduced to downloading the package, unpacking it, and executing the necessary script from the package.