We already have a cluster and instance of Aurora PostgreSql in abc region. Now as part of disaster recovery strategy, we are trying to create a read replica in a xyz region.
I was able to create it manually by clicking on "Add Region" in AWS web console. As explained here.
As part of it, following as been created.
1. A global database to the existing cluster
2. Secondary region cluster
3. Secondary region instance.
Everything is fine. Now I have to implement this through cloud formation script.
My first question is, can we do this through Cloud formation script without losing data if primary cluster and instance already created ?
If possible, please share aws doc for cloud formation scripts.
Please see the other post on this subject: CloudFormation templates for Global Aurora Database
The API that is required for setting up the GlobalCluster is AWS::RDS::GlobalCluster and this is currently not listed in CloudFormation documentation.
I was able to do the same using Terraform and that is documented for PostgreSQL here: Getting Aurora PostgreSQL Global Database setup using Terraform
Related
I have some Aurora PostgreSQL Clusters created on our AWS account. Because of some access issues (which we are working on already), there are several people in other teams who create random DB's on these Aurora Clusters and then we need to work on cleaning them up.
I wanted to check if there is a way to get alerted (via SNS Notifications etc.) whenever a new DB is created on these AWS Postgres clusters using some AWS Tools itself.
Thanks
You could do it using AWS Aurora Database Activity Streams, it will capture all database activity on the database and send it AWS Kinesis Data Stream and you could create a AWS Lambda function to read Kinesis Data Stream and identify the events needed (ex. create database)and finally send notification to AWS SNS from AWS Lambda code.
Another option is enable pgaudit on your AWS Aurora PostgreSQL, send logs to AWS CloudWatch, create AWS Lambda to read the events from AWS CloudWatch and send AWS Notification
Below you can find step by step on AWS Blog below.
Part 2: Audit Aurora PostgreSQL databases using Database Activity Streams and pgAudit
Are there any tools for database administration that can be deployed in AWS Lambda? My usecase is i've aurora serverless running inside a vpc and I want an AWS Lambda function to be able to visualize, clear and delete datas so developers do not need to get inside bastion hosts everytime they need to clear a row.
There is Data API for Aurora Serverless which allows you to use regular AWS SDK (e.g. boto3) to query your databases in Aurora Serverless.
I am looking for a way to manage schema changes to my AWS Aurora Postgres instance.
My whole AWS stack is set up using a Cloudformation template which is used to automatically deploy the stack when a change is detected in the source control. The Cloudformation template is built, a change set is prepared and finally excecuted on the stack.
I was hoping that the table definition of my Aurora instance could go inside the Cloudformation template somehow, so the schema migrations could be a part of the change set. Is this possible?
Note, I have seen this recommendation: https://aws.amazon.com/blogs/opensource/rds-code-change-deployment/
For anything custom like that use a Custom Resource Lambda that you can include in your Cloud Formation stack. The Lambda will need a layer for your postgress driver and it needs to include the migration script in the Lambda.
See the answer at this link, you will get 3 different options how you can trigger the Lambda.
Is it possible to trigger a lambda on creation from CloudFormation template
how to setup tables of aws aurora mysql using aws cloudformation or aws cdk?
in my setup i have a serverless app using lambda for various microservices. the datebase is a serverless aurora mysql database. to provision the aws infrastructure i will use aws CDK. i like to setup the database using some migration tools like liquibase or sequelize.
for the moment i am using a separat lambda function. the lambda function executes liquibase to setup db changes. but i have to execute the function separately after deployment with CDK is succeded.
an execution triggered after the execution of the cloudformation stack (cdk stack) would be optimal?! I would like to avoid a CI / CD stack via code pipeline.
does anyone has best practice to setup database at provision time?
Cloud watch rules
Cloud watch rules based on cloudformation events can be used to route events for processing lambda. Cloud watch rules can be a part of the CDK deployment description.
The triggered function can then execute liquibase, flyway, sequelize or something else to spinup or change db.
---- or ----
Cloudformation custom resource
AWS cloudformation custom ressource can execute a lambda function during cloudformation lifecycle.
The triggered function can then execute liquibase, flyway, sequelize or something else to spinup or change db.
I use Cloudformation custom resources for running database migrations and initial database setup scripts at deployment time.
This is the recommended way for running DB migrations for serverless applications if you don't want to rely on a CI/CD pipeline to do it for you.
Here's a well written blog post by Alex DeBrie about CF custom resources: https://www.alexdebrie.com/posts/cloudformation-custom-resources/
I use CloudFormation's AWS::RDS::DBCluster resource to create my Aurora MySQL database cluster.
My question is, has anyone created stored procedures as well as events in Aurora MySQL via CloudFormation? Is that even possible?
Delivering these via CloudFormation would allow me to recreate the infrastructure without deploying the stored procedures and events separately.
There's no way to configure stored procedures and events with the AWS::RDS::DBCluster CloudFormation resource directly.
My suggestion would be to provision an AWS::EC2::Instance containing a UserData script that installs the mysql client, then executes the contents of a user-provided MySQL script creating events/stored-procedures on the newly-created DB instance.