Twitter feed back end Use docker-compose and any programming language
- Implement an endpoint to add message
- Implement an endpoint to get feed (get existing messages and stream new ones - use HTTP streaming)
- Implement back pressure for message creation (use RabbitMQ/Kafka)
- Use Cockroachdb(at least 2-node cluster) as a database
- Implement a bot to generate messages (at configurable speed)
CRITICAL - Project must start with one command (bash file) without installing anything except docker Result is a link to a git project
To startup application use ./run.sh (CRITICAL requirement).
** If you see something like this in the console
ERROR: cluster has already been initialized
Failed running "init"
no change
startup was completed successfully.
Type
docker compose ps -a
to see your containers are running.
Type
docker compose logs -f
to see containers logs.
To stop application use ./stop.sh
Use Frontend repository for frontend implementation.
You can open multiple tabs and you will receive on each tab updates from bot / other users.
Bot'll begin creating messages after 15 seconds (configurable in docker-compose.yaml argument --delay for bot service command).
To increase bot's post frequency, increase --reqs-per-min argument for bot service command in docker-compose.yaml.
docker compose exec -it roach1 ./cockroach sql --host=roach1:26257 --insecure enter cockroach.
Look into docker-compose.yaml for better usage userstanding.
Entrypoints for containers are in cmd/*.
Three (bot, api, sse) containers are using same image, so you need to build it before (docker build -t twitter-feed .). This command is already in use in ./run.sh.
I decided divide endoints for list messsages and SSE fetching, because as for me, these endoints should be accepted by defferent containers, so in the future they may need independent scaling, configuration, etc.
- Better configuration
- Tests
- Graceful shutdown for all services
Please contact me if you have other questions.