README
¶
Adheretech Interview
Usage
You'll need to provide a host for the token source and credentials for the DB. The relavant variables are here. Once you've got those, you can run the following command to get a token:
docker run --env-file .env datawitch/at-token-service
This'll run the task with a default of 5 tokens. The container takes in the size arg via CMD though, so docker run --env-file .env datawitch/at-token-service 20 will get you 20 tokens. Any other positive integer up to 455902 (the maximum set by the token provider) works.
Environment variables
| Variable | Description |
|---|---|
TOKEN_HOST |
Host for the token source |
DB_HOST |
Host for the token DB |
DB_PORT |
Port for the token DB |
DB_USER |
User for the token DB |
DB_PASS |
Password for the token DB |
DB_NAME |
Name of the token DB |
DB_SSLMODE |
sslmode for the token DB. Defaults to require |
Local development
Included in this repo is a docker-compose file that can be used to run the service locally. It sets up a local PostgreSQL instance that'll reject tokens that include a -, and a token source that provides random strings for tokens. There isn't a wait-for-it.sh script, so you'll want to start the database service manually (docker-compose up -d database source) and then run docker-compose up app.
Future work
- Cloud Run. As mentioned above, this is a few small tweaks away from being a GCP Cloud Run service.
- Secrets management. Passing in creds via env vars comes with some security issues. Using a secret store like Vault or GCP can inject directly into running resources.
- Background tasks. If you request a large number of tokens from the service, network I/O is the limiting reagent (we're making a lot of DB calls to meet the "1-by-1" requirement). That's annoying, since CPU and memory are easier to scale, and the call blocks. Depending on how we're ultimately returning the results, it might be better to spin off a background call and return them asynchronously (say, a gRPC stream).
- CI/CD. Right now this is living on GitHub. GitHub Actions could be used to deploy to wherever (Docker Hub, GCP, etc). Other tools like CircleCI would work well too.
Problem Statement
- Obtain some string tokens from an HTTP server.
- Insert them 1-by-1 into a SQL database.
- Output each token and whether the insert was successful. Due to a database constraint, inserts will fail for some tokens.
- Include a Dockerfile to containerize your solution. We should be able to test the solution by building and running the container.
- Document the solution as a technical design proposal.
Documentation
Source Files