Python scripts to be run using AWS's Lambda service to Create and Delete Snapshots of EBS Volumes
I would recommend to start using Amazon Data Lifecycle Manager.
[Read my blog post for more details on setting this up in Lambda if you have not used it before.] (https://www.evergreenitco.com/evergreenit-blog/2016/4/19/aws-ebs-backup-job-run-by-lambda)
First create an IAM policy called "ebs-backup-worker" with the following policy document:
{
"Version": "2012-10-17",
"Statement": [
{
"Effect": "Allow",
"Action": [
"logs:*"
],
"Resource": "arn:aws:logs:*:*:*"
},
{
"Effect": "Allow",
"Action": "ec2:Describe*",
"Resource": "*"
},
{
"Effect": "Allow",
"Action": [
"ec2:CreateSnapshot",
"ec2:CopySnapshot",
"ec2:DeleteSnapshot",
"ec2:CreateTags",
"ec2:ModifySnapshotAttribute",
"ec2:ResetSnapshotAttribute"
],
"Resource": [
"*"
]
},
{
"Effect": "Allow",
"Action": [
"sns:Publish"
],
"Resource": "*"
}
]
}
Next create an IAM role also called "ebs-backup-worker" select "AWS Lambda" as the Role type, then attach the "ebs-backup-worker" policy created above. When completed and you check the trust relationship in the role through "Edit Trust Relationship" it should look like below:
{
"Version": "2012-10-17",
"Statement": [
{
"Effect": "Allow",
"Principal": {
"Service": "lambda.amazonaws.com"
},
"Action": "sts:AssumeRole"
}
]
}
Add the regions you want run the scripts against as a Python Base64 encoded string Lambda environment variable "aws_regions".
Since Lambda does not allow commas in the environment variable values, we cannot enter in a list for our regions we want to run the script against. To work around this we will Base64 encode the list/string, and then decode the string in our script and then "split" the string into a list again.
Below is an example of using Python to Base64 encode our string:
~$ python
Python 2.7.12 (default, Nov 19 2016, 06:48:10)
[GCC 5.4.0 20160609] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import base64
>>> encoded = base64.b64encode(b'us-west-2,us-east-2')
>>> encoded
'dXMtd2VzdC0yLHVzLWVhc3QtMg=='
>>> data = base64.b64decode(encoded)
>>> data
'us-west-2,us-east-2'
>>>
We will copy the encoded value and add it as the Lambda environment variable "aws_regions". When copying the encoded value please omit the single quotes in the output (ie. dXMtd2VzdC0yLHVzLWVhc3QtMg==).
This is optional environment variable if you want publish any topic, so you might receive email notification once backing up was executed.
Create two functions in Lambda using the Python 2.7 runtime, one for the backup script and one for the cleanup script. I recommend just using the 128 MB memory setting, and adjust the timeout to 10 seconds (longer in a larger environment). Set the event source to be "CloudWatch Events - Schedule" and set the Schedule expression to be a cron expression of your liking i.e. "cron(0 6 * * ? *)" if you want the job to be kicked off at 06:00 UTC, set the cleanup job to run a few minutes later.
Optionally a third function can be created using lambda-ebs-copy.py
to copy snapshots to a different region for increased redundancy. The env variable aws_copy_region
specifies the destination region of the copy.
You will need to tag your instances in order for them to be backed up, below are the tags that will be used by the Lambda function:
Tag Key | Tag Value | Notes |
---|---|---|
Backup | Value Not Needed | |
Retention | Number of Days to Retain Snapshot | Default is 7 Days |
Skip_Backup_Volumes | volume id(s) in CSV string | List either a single volume-id, or multiple volumes-ids in a Comma Separated Value String |
[Again if you need more details on setting this up please check my blog post.] (https://www.evergreenitco.com/evergreenit-blog/2016/4/19/aws-ebs-backup-job-run-by-lambda)