geekyme / fooddelivery-temporal

About this repo

This is a demo to show how temporal could be used to orchestrate a food delivery's actors - restaurant, courier, and the order

Prerequisites

Install docker

Setup

1. Build courier service:

./gradlew :courier:jibDockerBuild

2. Build restaurant service:

./gradlew :restaurant:jibDockerBuild

3. Build issuance service:

./gradlew :issuance:jibDockerBuild

4. Run full stack including temporal:

docker-compose up

Test running workflows

1. Go to issuance service on http://localhost:8082/graphiql, and enter the graph query to create an order:

mutation {
  fulfillOrder(order: { id: "my-test-order-id-1"}) {
    id
  }
}

2. Observe issuance logs has some entry

3. Observe in temporal UI on http://localhost:8088/namespaces/default/workflows?range=last-1-days that a workflow is started. Copy the workflow id you see for RestaurantOrderWorkflow.

4. Go to restaurant service on http://localhost:8081/graphiql, and enter the graph query to make restaurant accept order:

mutation {
  acceptOrder(order: { id: "<restaurant_order_workflow_id_from_earlier>"}) {
    id,
    status
  }
}

5. Observe restaurant logs has some entry

6. Observe in temporal UI on http://localhost:8088/namespaces/default/workflows?range=last-1-days that a new CourierOrderWorkflow is started. Copy the workflow id you see for CourierOrderWorkflow.

7. Go to restaurant service on http://localhost:8083/graphiql, and enter the graph query to make courier accept order:

mutation {
  courierAcceptOrder(order: { id: "<courier_order_workflow_id_from_earlier>"}) {
    id,
    status
  }
}

8. Observe courier logs has some entry

9. Temporal UI should show master workflow (IssuanceWorkflow) and child workflows (CourierOrderWorkflow and RestaurantOrderWorkflow) are now completed

Development

1. Kill the service in docker, for example courier:

docker-compose stop courier

2. Run application manually outside docker:

./gradlew :courier:bootRun

Monitoring for temporal

The local setup is using grafana for data visualizaton and prometheus for data collection. In production, we will most likely use DataDog / AWS Cloudwatch Metrics.

Temporal allows customization of metrics endpoints. For this example, we configured temporal's PROMETHEUS_ENDPOINT. See the docker-compose.yml file for more details.

After that, the stats can be monitored on localhost:3000.

Running loadtest

1. Install k6

2. Go to loadtest/ and run k6 run loadtest.js. You may edit the code as you wish