AWS EC2 Volume manager

The ec2-volume-manager, manages the attachment of volumes to ec2 instances. When the instance is stopped or terminated, the volume attachments are removed. When a new instance is started, an volume is attached to it. The goal of the volume manager is to be able to update the AMI of a stateful server, without loosing the data by defining a autoscaling group of instance size 1 and perform a rolling update.

How does it work?

The manager will listen to all EC2 instance state change notifications. When an instance with the tag ec2-volume-manager-attachments reaches the state running, it will attach all volumes with the same tag and value. The volumes will need to have a tag device-name which is used in the volume attachment.

How do I use it?

You can start using the EC2 volume manager, in three simple steps:

1. deploy the ec2-volume-manager
2. create and tag the volumes
3. create an auto scaling group size 1, propagating the tag

deploy the ec2-volume-manager

To deploy the provider, type:

git clone https://github.com/binxio/ec2-volume-manager.git
cd ec2-ec2-volume-manager
aws cloudformation deploy \
        --capabilities CAPABILITY_IAM \
        --stack-name ec2-volume-manager \
        --template ./cloudformation/ec2-volume-manager.yaml

Create one or more volumes to attach

Create the volumes to attach and tag them with an ec2-volume-manager-attachment value:

  Disk1:
    Type: AWS::EC2::Volume
    Properties:
      AvailabilityZone: !Sub '${AWS::Region}a'
      Size: 8
      Tags:
        - Key: ec2-volume-manager-attachment'
          Value: stateful-instance-1
        - Key: device-name
          Value: xvdf

  Disk2:
    Type: AWS::EC2::Volume
    Properties:
      AvailabilityZone: !Sub '${AWS::Region}b'
      Size: 8
      Tags:
        - Key: ec2-volume-manager-attachment
          Value: stateful-instance-1
        - Key: device-name
          Value: xvdg

Create an auto scaling group

Create an auto scaling group and apply the tag ec2-volume-manager-attachment to the instances:

  AutoScalingGroup:
    Type: AWS::AutoScaling::AutoScalingGroup
    Properties:
      MinSize: '0'
      MaxSize: '1'
      DesiredCapacity: '1'
      Tags:
        - Key: ec2-volume-manager-attachment
          Value: stateful-instance-1
          PropagateAtLaunch: true
    UpdatePolicy:
      AutoScalingRollingUpdate:
        MinInstancesInService: 0
        MaxBatchSize: 1
        WaitOnResourceSignals: true

That is all. If you want to see it all in action, deploy the demo.

Note that the instance should wait in the boot script, until all volumes have been attached before proceeding with the boot sequence and report succesful completion using cfn-signal.

Deploy the demo

In order to deploy the demo, type:

export VPC_ID=$(aws ec2  --output text --query 'Vpcs[?IsDefault].VpcId' describe-vpcs)
export SUBNET_IDS=$(aws ec2 describe-subnets --output text \
                --filters Name=vpc-id,Values=$(VPC_ID) \
				          Name=default-for-az,Values=true \
                --query 'join(`,`,sort_by(Subnets[?MapPublicIpOnLaunch], &AvailabilityZone)[*].SubnetId)')
aws cloudformation deploy \
        --capabilities CAPABILITY_NAMED_IAM \
        --stack-name ec2-volume-manager-demo \
        --template ./cloudformation/demo-stack.yaml \
        --parameter-overrides VPC=$VPC_ID Subnets=$SUBNET_IDS

Read the blog too: How to update an ec2 instance with volume attachments using CloudFormation