0

I'm fairly new to AWS CDK through code and I've been looking around how to restore a database. I understand that you cannot restore Aurora database into an existing instance from snapshot (1) (2). My issue lies in the fact that we use infrastructure as a code, and therefore I need to sync the "new" database into the "old" IaC (e.g. number of instances is different)

I have managed to create a new database and change all the configurations required. However, one point I'm missing is aligning this with aws-cdk. Specifically:

  • I create a new database cluster from and old snapshot
  • I change the connections of my application to use this new database cluster
  • I sync my IaC with the AWS state

The last point is the one that is eluding me. How do I set up my code using aws-cdk so that I use the restored database (setup, migrations, instances, ...) rather than the old one?

I've also tried to change the name to the restored one. However, I then encounter issues when I deploy to different environments - they create a new database rather than work with the existing one already.

Improve this question

1 Answer 1

Reset to default
2
+50

In your CDK app, remove the existing cluster resource and add a new one. The "adding" step can be done on a snapshot or an existing cluster.

From a snapshot

CDK and CloudFormation can create a new cluster from a snapshot. The relevant CDK construct is DatabaseClusterFromSnapshot.

The steps would be:

  1. Remove the existing cluster construct from the stack. Pay attention to the cluster's removal policy, which governs what happens to a removed resource. Deploy the changed app to remove the "old" cluster from CDK management.
  2. Create a new cluster from a snapshot. Pass a snapshot identifier (name or ARN) as a prop to DatabaseClusterFromSnapshot.

From an existing cluster

Alternatively, you can import an existing cluster into a CDK stack with the cdk import CLI command. CDK import uses CloudFormation import functionality under the hood. AWS::RDS::DBCluster is a resource type that supports import.

Once you have imported the cluster, it will be part of your CDK stack.

The steps would be:

  1. Step 1 as above
  2. Import the cluster by adding new cluster construct, following the step-by-step instructions for cdk import. Note the doc's warning Do not add any other changes! You must also make sure to exactly model the state that the resource currently has.
Improve this answer
Sign up to request clarification or add additional context in comments.

2 Comments

Thanks a lot. I've accepted the answer as it does indeed do what I want to do - DatabaseClusterFromSnapshot does it for me. I have two more question - Can I delete the snapshot afterwards, or do I need to keep the old snapshot for the entire life of the cluster created from such snapshot? And what if we use CDK to support multiple environments (test/prod) and the snapshot needs to be restored only on one of them? Note that I'm trying to work with the automated backup functionality which is removed after a day if I understand the docs correctly. My apologies if it doesn't belong here.
@SimProch Glad to help. I don't think your snapshoot is required once you have restored it. Your CDK code should keep the snapshot ID prop, though. Otherwise the CDK will think it needs to perform an update on a new or blank snapshot. Consider conditional logic like if (env === 'prod') to get the differential behaviour by environment.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.