isabella232 / auto-snapshot

Creating snapshots for selected EBS volumes in AWS

humilis-autosnapshot

Creates scheduled snapshots of selected EBS volumes in the AWS cloud.

NOTE: This repo has been deprecated by the more comprehensive EBS snapshot scheduler supported by AWS.

How does it work?

Once deployed in the AWS cloud, the system creates snapshots from every EBS volume that has the following tags:

tag key                             tag value
-----------------------------------------------------------------------------------
*Name                               <volume friendly name>
*auto:snapshots                     <any value>
auto:snapshot:topic                 <topic name for SNS event>
auto:snapshots:retention_days       <number of days to keep the snapshot>

*mandatory tags

Example tags:

tag key                             tag value
-----------------------------------------------------------------------------------
Name                                my-super-important-volume
auto:snapshots                      yes
auto:snapshot:topic                 snapshot-was-created-for-my-super-important-volume
auto:snapshots:retention_days       15

Development

Assuming you have virtualenv installed:

cd lambda
make develop

. .env/bin/activate

To run the test suite:

make test

Deployment

Note: if you don't know what humilis it, it is strongly recommended that you take a quick look at the docs before reading further.

First you will need to install and configure humilis, if you haven't do so already. Then you can make a test deployment to AWS using:

make create

which will deploy in stage TEST a dummy environment that contains just the autosnapshot layer. You can then go to the AWS Lambda console and test the deployment there.

In a real deployment the autosnapshot layer is unlikely to stand on its own, but to be part of a larger environment. Your environment definition file may look something like this:

my-environment:
    description:
        Everything I need to keep my company running
    layers:
        # The vpc layer takes care of setting the networking stuff
        - {layer: vpc}
        # Deploy a layer with some RDS instances
        - {layer: databases}
        # Deploy a layer with some EC2 instances
        - {layer: instances}
        # Deploy the autosnapshot layer
        - {layer: authosnapshot}

This means that your environment directory tree may look something like this:

.
├── layers
│   ├── autosnapshots
│   │   ├── meta.yaml
│   │   ├── outputs.json.j2
│   │   └── resources.json.j2
│   ├── dababases
│   ├── instances
│   └── vpc
└── myenvironment.yaml

where the layers/autosnapshots directory has the same contents as this repository. The best approach to do this is to add this repository as a subtree of your environment repository. I recommend you use stree to manage git subtrees:

# Assume you are in the root directory of your environment, i.e. where the
# environment definition file is located.
git stree add autosnapshot -P layers/autosnapshot \
    git://github.com/Innovativetravel/auto-snapshot.git

Once the autosnapshot subtree has been added to your environment repository, you would deploy the snapshot layer as part of the deployment of your whole environment. E.g. by running a command similar to this one:

humilis --profile [profile] create --stage [stage] myenvironment.yaml

where [profile] is the humilis configuration profile (which determines, e.g. which AWS credentials should be used for deployment), and [stage] may be used when you want to have multiple deployments of your environment running side by side (e.g. by deploying first to stage DEV and then to stage PROD).