Skip to content

Latest commit

 

History

History
187 lines (141 loc) · 4.41 KB

week10.md

File metadata and controls

187 lines (141 loc) · 4.41 KB

Week 10 & 11 — CloudFormation

Quick Links

Weekly Outcome

Week 10

Week 11

Weekly Summary

Reflection

Challenges

I had a lot of difficulty deleting the ALB and the ALB Security Group from CloudFormation due to the tight coupling of the cluster stack with the networking one's subnets. The network interfaces were in use. I had to wait for the deletion to finally fail then manually delete the ALB. After it was deleted I had to wait for the interfaces to go from "in use" to available.

Knowledge Transfer

CloudFormation Syntax

In CloudFormation templates:

YAML compiles down to JSON so,

VPC: 
# ...

IGW:
  Type: AWS::EC2::InternetGateway
  Properties:
    VpcId:
      Ref: VPC

is equivalent to

VPC: 
# ...

IGW:
  Type: AWS::EC2::InternetGateway
  Properties: {VpcId: {Ref: "VPC"}}

Functions

You can also use CloudFormation's various functions to reference previously defined resources using Ref, GetAtt, etc

VPC: 
# ...

IGW:
  Type: AWS::EC2::InternetGateway
  Properties:
    Tags:
      - Key: Name
        Value: CruddurVPC

AttachIGW:
  Type: AWS::EC2::VPCGatewayAttachment
  Properties:
    VpcId:
    # Get ID from VPC 
      Ref: VPC
    # Get ID of previously defined Resource "IGW" using CloudFormation function `!Ref`
    InternetGatewayId: !Ref IGW

Converting Manually Deployed Services

service-backend-flask.json

  "serviceConnectConfiguration": {
    "enabled": true,
    "namespace": "cruddur",
    "services": [
      {
        "portName": "backend-flask",
        "discoveryName": "backend-flask",
        "clientAliases": [{"port": 4567}]
      }
    ]
  }

ServiceConnectConfiguration:
  Enabled: true
  Namespace: "cruddur"
  # TODO - If you want to log
  # LogConfiguration
  Services:
    - DiscoveryName: backend-flask
      PortName: backend-flask
      ClientAliases:
        - Port: !Ref ContainerPort

Key Takeaways

  • Check your Network ACLs have outbound routes if you're having issues
  • Update Behavior, Modification vs Replacement, in CloudFormation causes a lot of unexpected issues
  • Logical Names for resources have a length limit, so try to keep them short including when using !Sub to change them
  • Dependencies of resources such as networking security groups makes it difficult to reduce coupling in CloudFormation

Questions

Required Homework

Deploy CloudFormation

Cluster

Delete previously manually configured:

  • ALB (under EC2)
  • Target Groups (under EC2)
  • ECS cluster 'cruddur' and its services
  • Cloud Map namespace 'cruddur'

RDS

After Deployment:

  • update connection url var in parameter store

Service

Delete previously manually configured:

  • IAM Role: CruddurServiceExecutionRole

Note: Health check fails if database is not on the same VPC

Configure:

  • add load balancer dualstack.CrdClusterALB to Route53 "A" records, api.domain and domain

Fixes:

  • HealthCheckPort in cluster template should be set to 4567, not 80.

Port for health check wasn't updating in target group after update to changeset, so I deleted all the stacks.

DynamoDB

Delete previously manually configured:

  • log group: /aws/lambda/cruddur-messaging-stream
sam build \
--use-container \
--config-file $CONFIG_PATH \
--template $TEMPLATE_PATH \
--base-dir $LAMBDA_DIR

Note:

  • --base-dir is where the sam command is executed, so in the template.yaml, the CodeUri is a folder relative to this path
  • e.g.) for the folder ./aws/lambdas/cruddur-messaging-stream in the workspace:
    • --base-dir $ABS_PATH/aws/lambdas/
    • CodeUri: cruddur-messaging-stream

CICD

after deployment:

  • codepipeline > settings > "pending connection" > update connection

Frontend

Delete previously manually configured:

  • Route 53 > Hosted Zone > delete A Record for kapingkaping.cloud

After deployment:

  • build static site
  • upload build/ contents into kapingkaping.cloud S3 bucket
    • make sure the static/ sub-directory is uploaded in addition to the files