k8s-rds-autoscaler
make RDS Aurora Serverless cluster autoscaling react quicker by scaling it actively from K8s
Usage
Step 1
Make sure there is a common label selector for the group of pods you are going apply RDS autoscaler.
The label "app.kubernetes.io/instance" can be used by Helm convention.
Step 2
Specify pod connection usage via rds-autoscaler/connections-usage pod annotations.
Your pod specification should look like this:
apiVersion: v1
kind: Pod
metadata:
annotations:
rds-autoscaler/connections-usage: 100
...
Step 3
Get an AWS access key and make sure it is allowed to perform rds:ModifyCurrentDBClusterCapacity operation over your RDS cluster.
Step 4
Deploy the autoscaler via Helm
BUILD_TAG=$(date +%s); \
IMAGE_URL="<MY_DOCKER_REPO_URL>"; \
docker build -t $IMAGE_URL:$BUILD_TAG /<PATH_TO>/k8s-rds-autoscaler && \
docker push $IMAGE_URL:$BUILD_TAG && \
helm upgrade -i rds-autoscaler /<PATH_TO>/k8s-rds-autoscaler/deployments/helm/rds-autoscaler -n <MY_NAMESPACE> \
--set image.repository=$IMAGE_URL,image.tag=$BUILD_TAG \
--set autoscaler.podsLabelSelector="<MY_EXAMPLE_LABEL_1>=<MY_EXAMPLE_LABEL_VALUE_1>,<MY_EXAMPLE_LABEL_2>=<MY_EXAMPLE_LABEL_VALUE_2>" \
--set-string autoscaler.connectionsHeadroom="50" \
--set autoscaler.scalingsSet='[{"scale":"2"\,"limit":157}\,{"scale":"4"\,"limit":315}\,{"scale":"8"\,"limit":631}\,{"scale":"16"\,"limit":1263}\,{"scale":"32"\,"limit":2526}\,{"scale":"64"\,"limit":4816}\,{"scale":"192"\,"limit":15160}\,{"scale":"384"\,"limit":30320}]' \
--set-string autoscaler.operationTimeout="10" \
--set autoscaler.dbIdentifier="<MY_RDS_INSTANCE_IDENTIFIER>" \
--set-string autoscaler.downscaleForbiddenWindowSeconds="300" \
--set-string autoscaler.syncIntervalSeconds="3" \
--set autoscaler.awsAccessKeyId="<MY_AWS_ACCESS_KEY_ID>" \
--set autoscaler.awsSecretAccessKey="<MY_AWS_SECRET_KEY>" \
--set autoscaler.awsRegion="<AWS_REGION>"
Step 4
Use kubectl logs to check if the autoscaler is running correctly.