This is a Go Project that should be used to create a cloud environment. The app create volumes, ec2 instances, security groups, etc, on AWS.

The easiest way to use cloud-machine is using Docker and the available cloud-machine image issuing the following commands:

$ docker pull neowaylabs/cloud-machine:latest
$ docker run --rm -it -v <local-config-dir>:/data neowaylabs/cloud-machine:latest /machine-up -h

The <local-config-dir> is some directory containing your infrastructure configurations (see section [How use?]). This place will be mapped to directory /data inside container.

Then, a simple usage could be:

$ docker run --rm -it -v <local-config-dir>:/data \
    neowaylabs/cloud-machine:latest \
    /machine-up /data/mongo-node.yml

To build cloud-machine you'll need Go >= 1.4 or use a docker image with Go installed. You can choose one of the following commands to build:

make build        # build using installed Go
make build-docker # build using docker

After that, two binaries will be created:

$ ./cmd/machine-up/machine-up --help
Usage of ./cmd/machine-up/machine-up:
  -access-key string
    	AWS Access Key
  -secret-key string
    	AWS Secret Key

$ ./cmd/cluster-up/cluster-up --help
Usage of ./cmd/cluster-up/cluster-up:
  -access-key string
    	AWS Access Key
  -secret-key string
    	AWS Secret Key

If you have Go installed, make install will install the binaries inside $GOPATH/bin.

On OSX the binaries will not work if built with docker because the image they were compiled in is an ubuntu. The recommended way in this case are installing Go or issuing the commands directly inside the development docker image (make shell-debug)

We have two different executables:

• machine-up: it's to create only one machine that have ONLY one instance and how many volumes you need.

• cluster-up: it's to create a cluster of machines, you need to tell it which machine-config use and how many of this machine should run.

IMPORTANT: Each machine will verify if you are creating new volumes, if yes a new provisory machine will be create only to format these volumes, after format the machine will be automatically destroyed. Cost will be applied.

This app will create a machine, it have a instance with one or more volumes, you need pass to it a machine-config file. You can have two section, one to describe information about your instance and another one with information about your volumes.

Instance obligatory parameters:

• name: It will be create a tag with Name key
• type: The type of instance to create
• imageid: The Image Id used to create the instance
• region: Region used to create the instance
• keyname: Key name used to permit access to instance
• securitygroups: Array of security group Id
• subnetid: Subnet Id of instance

Instance optional parameters:

• id: The id to load a already created instance. If you pass this property all other properties will be ignored
• availablezone: The available zone to create the instance and volumes
• cloudconfig: File that will be used to pass as userdata to instance
• ebsoptimized: If instance should be EBS Optimized, default is false
• shutdownbehavior: When you shutdown the machine will terminate or stop, default is stop
• enableapitermination: If you authorize terminate this instance by aws console, cli, etc, default is false
• tags: You can pass a list of key=values to add these tags to your instance

Volume obligatory parameters:

• name: It will be create a tag with Name key
• type: The type of volume to create, can be standard, io1 and gp2
• size: The size of volume to created
• device: The device used of this volume
• mount: Where should mount the volume
• filesystem: File system used to mount the device

Volume optional parameters:

• id: The id to load a already created instance. If you pass this property all other properties will be ignored
• snapshotid: When informed, the volume is created from an existing snapshot. In this case the volume is not formatted, obviously
• iops: The IOPS used to create volume, only to io1 type
• tags: You can pass a list of key=values to add as tags to your volume

IMPORTANT: If you have new volumes (without ID property or snapshotId) a new machine will be created only to format this volume, after format the machine will be automatically destroyed. Cost will be applied.

IMPORTANT: If you want to create a volume from a snapshot and increase the size of the new volume, you need to run a resize2fs as Aws Increase Volumes

Here we have an example of machine-config

# cloud-machine/mongo-node.yml
instance:
  name: mongo-node
  type: r3.xlarge
  imageid: ami-5d4d486d
  region: us-west-2
  availablezone: us-west-2a
  keyname: aws-keyname
  securitygroups: [sg-00000000,sg-00000001]
  subnetid: subnet-abcd0000
  cloudconfig: cloud-config/mongo-node.yaml
  ebsoptimized: true
  shutdownbehavior: stop
  enableapitermination: false
  tags:
    - { key: instancekey1, value: instancevalue1 }
    - { key: instancekey2, value: instancevalue2 }
    - { key: instancekey3, value: instancevalue3 }

volumes:
  - name: mongo-data
    type: io1
    size: 200
    iops: 1000
    device: /dev/xvdk
    mount: /data
    filesystem: ext4
    tags:
      - { key: volumekey1, value: volumevalue1 }
      - { key: volumekey2, value: volumevalue2 }
      - { key: volumekey3, value: volumevalue3 }

  - name: mongo-journal
    #id: vol-123456
    type: io1
    size: 25
    iops: 250
    device: /dev/xvdl
    mount: /journal
    filesystem: ext4

To run you need export AWS_ACCESS_KEY and AWS_SECRET_KEY or type aws configure and pass machine-config file

export AWS_ACCESS_KEY=<your access key>
export AWS_SECRET_KEY=<your secret key>

#OR

aws configure

./machine-up ./cloud-machine/mongo-node.yml

This app will create a cluster of machine, each machine is defined above, you need pass a cloud-config, that it's a file that you need to describe which machine-config.yml use and how nodes of this machine should run.

In cluster-config you can pass follow properties:

• machine: The file used to describe machine (see above)
• nodes: How many machines should be create

Sometimes you need use some default value to all your instances, for that leave theses fields empty inside of your machine spec, and fill inside of your default cloud spec. This is very helpful when you need update you image id or if you want create your infra in another region for example, in this case you only need update in one file.

Here we have an example of cluster-config

# cloud-machine/app-cluster.yml
default:
  imageid: ami-5d4d486d
  region: us-west-2
  keyname: aws-keyname
  securitygroups: [sg-00000000,sg-00000001]
  subnetid: subnet-abcd0000
  availablezone: us-west-2a
  tags:
    - { key: volumeAndInstanceKey1, value: volumeAndInstanceValue1 }
    - { key: volumeAndInstanceKey2, value: volumeAndInstanceValue2 }
    - { key: volumeAndInstanceKey3, value: volumeAndInstanceValue3 }

clusters:
  - machine: cloud-machine/mongo-node.yml
    nodes: 3

  - machine: cloud-machine/elasticsearch-node.yml
    nodes: 2

To run you need export AWS_ACCESS_KEY and AWS_SECRET_KEY or type aws configure and pass cluster-config file

export AWS_ACCESS_KEY=<your access key>
export AWS_SECRET_KEY=<your secret key>

#OR

aws configure

cluster-up ./cloud-machine/app-cluster.yml

If you have the permissions and are logged (using docker login) just run:

make publish

And a latest image will be generated. If you want to generate a specific version just run:

make publish version=<version>

Where version is the version that you are generating (should match with the tag, you created a tag... right ?)