We should set up GitOps for TinyHats to have automated CI/CD deployments of docker images for each microservice linked to each repo. Right now we have to manually build shit which isn't good.

Closes #73

Locust Load Generator with 2 parts

Custom requests (custom-locust-service)

Access test.tinyhat.me to specify parameters for testing. Host value must be http://gateway-service/ and a hat name. *Example: http://gateway-service/cat-ears

Note: Need to route test.tinyhat.me to custom-locust-service Load Balancer

Default requests (locust-service)

There is always a locust server running in the background to generate at least a 1 RPS (request per second) rate. The rate may fall below this level depending on the server response rate. Requests are randomly made to different hats.

Set up Staging Environment that deploys from staging branch

• [x] Spin down all existing EKS instances
• [x] Create EKS cluster called Tinyhats-Staging
• [x] Create Github Actions to:
  • [x] Automatically create docker image staging-add-service
  • [x] Automatically create docker image staging-admin-service
  • [x] Automatically create docker image staging-locust-service
  • [x] Automatically create docker image staging-fetch-service
  • [x] Automatically create docker image staging-frontend-service
  • [x] Automatically create docker image staging-manipulation-service
  • [x] Automatically create docker image staging-moderate-service
  • [x] Automatically create docker image staging-upload-service
• [x] Create Github Actions to automatically redeploy kubernetes pod when there is a codechange
• [ ] Set up Route53 to route staging.tinyhat.me to the staging environment.

Instead of having to switch docker images for the different scenarios, we should be able to have feature flags that can turn on and off various bugs that we have programmed in.

We need to add Kafka functionality to the application. To do this successfully, we need to add modifications

Pre-reqs

• [x] We need to make sure that each message for the Kafka Broker does not exceed 1MB. Therefore, the Kafka message CANNOT contain the bare base64 image. We need to change the code so with the image name or ID we can fetch the base64 data from the MySQL database for /manipulate and /fetch

Work

• [ ] Spin up Kafka broker namespace using Strimzi.
• [ ] Turn /fetch microservice into a Kafka producer, writing to the Kafka broker instead of sending a REST call
• [ ] Turn /manipulate microservice into a Kafka consumer, reading from the Kafka broker instead of receiving a REST call
• [ ] Enable Horizontal Autoscaling for /manipulate using this walkthrough

The email sent from /email will contain two links: one for approval and one for rejection. Each link will hit the moderate endpoint.

Moderation Service

Create a microservice that

• adds the user-submitted image to the database upon admin request

About

Accepts input in form-data, uploads to AWS S3 bucket, and pushes to MySQL AWS RDS. Generates random names for images to avoid conflict.

Closes #7 Closes #8

DB Structure

{"id":1,"keyId":"lfwy1ted33t35cnx","url":"https://uwuaascat.s3.amazonaws.com/lfwy1ted33t35cnx.jpeg","fileName":"lfwy1ted33t35cnx.jpeg","description":"baby lying down","approve":"false"}

Database Name: main Table name: images

Input

Uses multipart/form-data. image: file upload name: description of the image

Output

{
    "key": "3lh9ypagvkm2j4mh",
    "fileName": "3lh9ypagvkm2j4mh.jpeg",
    "url": "https://uwuaascat.s3.amazonaws.com/3lh9ypagvkm2j4mh.jpeg",
    "description": "baby lying down",
    "approve": "false"
}

Upload Service

Create a service that uploads an image to AWS S3, retrieves the S3 link, and stores it in AWS RDS. #7

API input

{
   image: `data`,
   category: "category",
   name: "catfish"
}

Reference readings

null

Spin up a https://github.com/typicode/lowdb instance to store all of the links to images and cat

{
       category: "cat",
       link: "awslink.com/s3",
       name: "Cat in the rain", 
}

• [x] Allow user to change model
• [x] Allow user to choose model
• [x] Make requests come from server, not client
• [x] Style new interface
• [x] Fix checkout cart

Closes #102

• accepts POST request to add hat onto image /manipulate
• accepts POST request to add new hat (interacts with MySQL to store and create preview images) /add

Scenario #1 - Bad Architecture

Services Changed: Frontend Service Problem: The frontend calls the manipulate service for every single hat at page load time instead of using the pre-manipulated images from the database. Folder: scenarios/bad-architecture/

Scenario #2 - Redis Memory Issues

Services Changed: YAML for Redis database Problem: We have allocated too little memory for redis to function properly Folder: scenarios/redis-memory/

Scenario #3 - Bad SQL Request

Services Changed: Product Catalog Service Problem: The SQL query to fetch hats is case sensitive. There are some images that break because of of the case sensitivity. Folder: scenarios/bad-sql/

**Problem: ** For many demo environments, it is difficult to execute code changes. **Solution: ** Create a helm chart that allows you to pull and execute various docker images with various environmental variables to make demo-ing easier.