erwijet / api

Central API managing access to DB and handling bagelbot slack slash commands

BagelBot 🥯

---

• BagelBot 🥯
  • Important External Resources
• Enviornment Variable Requirements
  • Tech Stack
  • Endpoint Mappings
  • Project Structure
  • Deployment Steps for bagelbot.erwijet.com

---

Important External Resources

• Bagelbot GitLab
• Bagelbot Deployment
• Bagelbot Slack Management
• Slack Visual Message Builder
• Bagelbot Container Registry
• Bagelbot Deployment Management
• Recommended DB Viewer
  • or use brew install --cask mongodb-compass if on a mac with brew.sh installed

Enviornment Variable Requirements

Running an instance of bagelbotdev/api requires the following enviornment variables specified, typically in a top-level .env file.

Name	Purpose
MONGO_URL	The mongodb connection string with credentials specified
COIN_ENDPOINT	The coin api endpoint to interact with. Typically, this is http://bryxcoin01.csh.rit.edu
SLACKBOT_WEBHOOK	The webhook to interact with when sending messages to slack channels

Slack @tyler with any questions or for access credentials.

Tech Stack

Sector	Technology	Justification
API Language	typescript	Type safety and seamless integration with other technologies in this project
API Framework	express	Middleware-based routing model, plus general familiarity
Bagelbot Database	mongodb	Mostly to spite @galen.guyer, but also familiarity with the mongoose modeling tool. Bagelbot implements a fixed-schema data set, despite mongodb.
GQL Client	urql	Lightweight alternative to apollo. Limited gql functionality is needed for this project

Endpoint Mappings

Endpoint	Description
GET /	Redirects to the #bagels channel in bryx slack
GET /kube	Redirects to the kubernetes dashboard for bagelbot. Note that this requires kubectl proxy to be running on the developer's machine, as well as a ~/.kube/config that grants access to the underlying deployed api instance.
GET /dashboard	Redirects to the bagelbot dashboard
GET /healthcheck	Flexible endpoint designed for confirming a 200 response
POST /interaction	Handles any slack interaction performed (such as clicking buttons in forms, etc)
POST /event	Handles scoped slack events performed (such as @BagelBot mentions, etc)
POST /slash/<command>	Handles the corresponding slash command, such as /register, /pay, etc
GET /v1/users	Returns the first name, last name, and slack id of all registered users
GET /v1/tabs	Get the opener's fullname, opened at timestamp, cartGuid, and closed status of all past and current tabs
GET /v1/coin	Returns the first name, last name, and coin balance for each registed user

Project Structure

Path	Purpose
@types/	Exposes non-scoped type declarations to be used anywhere in the project
gql/	Prebuilt GQL queries for fetching data from the Balsam Bagel ToastTab GQL endpoint
src/balsam	Logic relating to interacting with Balsam Bagels (and their gql endpoint)
src/db	Models and Schemas, as well as logic wrappers for interacting with the Bagelbot DB
src/coin	Logic for interacting with the underlying coin blockchain
src/db/models	Mongoose model definitions. Should contain a one-to-one file mapping with the schemas folder
src/db/schemas	Mongoose schema definitions, for consumption by the models folder
src/middlewares	Express middlewares, such as validating registration, etc
src/routes	Contains all the routers to be used by the express app. Each folder / file within this directory correspond to a part of the path to be handled. For example, src/routes/slash/menu.ts should be set up to handle the /slash/menu request, and so on.
slack/	Resources for interacting with Slack
slack/blockkit/prefab	Static blockkit definitions. See the Slack Visual Message Builder for context
slack/blockkit/mappers	Functions that, given a fixed input, returns a blockkit message definition. These are the dynamic counterparts to the prefab messages

Deployment Steps for bagelbot.erwijet.com

The production enviornment is not tied to github CI/CD. The purpose of this, is that often times chagnes need to be tested before committing. In the furture, we may have a speerate bot that acts as a "staging" enviornment, but it is nearly impossible to replicate a staging slack enviornment for development. As such, the production enviornment acts as a staging enviornment.

Production deployment prerequisites:

1. The yarn, kubectl, and docker cli tools are all installed and accessable in $PATH.
2. Your local container runtime is up.
3. You have access to the underlying kubernetes cluster (specifed in your local ~/.kube/config)
4. You are logged in to the relevant container registry. You can do this with docker login ghcr.io/bagelbotdev

Once you have confirmed you meet the prereqs, you can deploy the current working version of the codebase to production with yarn rollout which will automatically perform the following steps:

1. Build the current version of the codebase into an image tagged with :latest.
2. Push the newly created image to the container registry.
3. Restartes the kubernetes api deployment which will trigger an image re-pull.
4. Executes a healthcheck probe twice a second to check when the deployment is back online. When the response changes from 404 not found or Bad Gateway to OK, the deployment is complete!