A utility AWS lambda function to manage EBS and RDS snapshot backups.
The Lambda function takes new backups when executed, and manages the deletion of the old ones when the upper limit is reached.
The origin of this project comes from: https://github.com/evannuil/aws-snapshot-tool
The tool uses the supplied boto3
library to connect to the AWS account, and uses the IAM Role defined in the CloudFormation stack to enable access to the required assets.
The only external dependency is on the pytz
library for some basic time processing functions.
To deploy the lambda code you can run the supplied upload_lambda.sh
script, or modify it for your purposes.
The lambda deployment process requires an s3 bucket to store the code before deployment, so before running the script you need create the bucket and export a BUCKET
environment variable for the script to use.
for example:
export REGION=ap-southeast-2
export BUCKET="lambdabucket1978"
aws s3 mb s3://${BUCKET} --region ${REGION}
Once the BUCKET
and optional REGION
variables are set, when you run upload_lambda.sh
it will do the following:
- Use
pip
to install thepytz
library alongside the python lambda function - Use
cloudformation package
to zip up the application and upload to s3 - As part of the
cloudformation package
, a newgenerated-cloudformation.yaml
file will be created with theCodeUri
pointing at the newly uploaded zip file - Invoke a
cloudformation deploy
to execute the creation of a new stack namedaws-backup-lambda
Once complete, you should have a new CloudFormation stack, which will have created the Lambda and all required AWS assets
The configuration of the lambda is detailed in the cloudformation.yaml
template file.
Things you might want to review and change are:
- The frequency of the backups
- The tags used for the snapshots
- The date based labels to use for each snapshot
- The number of snapshots to keep
- Enable / Disable the EBS or RDS backup function
The configuration for the Lambda is managed as the Input
passed to the function from the Scheduled event trigger.
An example configuration might be:
{
"period_label": "day",
"period_format": "%a",
"keep_count": 14,
"ec2_region_name": "ap-southeast-2",
"rds_region_name": "ap-southeast-2",
"tag_name": "MakeSnapshot",
"tag_value": "True"
}
period_label
is used to identify all backups in the same set, ensure this is UNIQUE across each scheduled eventperiod_format
is the format of the current time to apply to each of the backups, more detail belowkeep_count
the number of snapshots to keep for eachperiod_label
ec2_region_name
if supplied, EBS volumes for the specified region will be included in the backup runrds_region_name
if supplied, RDS instances for the specified region will be included in the backup runtag_name
the RDS and EBS items need to have this tag name to be considered part of the backuptag_value
the RDS and EBS items need to have this tag value to be considered part of the backup
Both EBS and RDS Snapshot management is supported and enabled by default.
Control of which service is executed is identified by the suppling the region name for the service.
- Supply
ec2_region_name
to run the EBS snapshot process - Supply
rds_region_name
to run the RDS snapshot process
Note: Currently the backup function only runs against a single region, you could easily add another copy of the function to run against an additional region.
Note: For RDS cluster support just add tag_name
, tag_value
to one of the instance in the cluster
Both of these services will use the same field for tag_name
, tag_value
and the keep_count
fields, if you need them to differ then create another Event trigger with different parameters.
The syntax used to label the backups with a value indicating the current time of the backup.
Any python time format string will be supported, with a set of suggested values shown below.
Suggested settings:
- More often than a day:
%a%H
- show the day of the week and the hour of the day - Daily:
%a
- show the day of the week - Weekly:
%U
- show the week of the year - Monthly:
%b
- show the month