TCP server with rate limit within partition.
Partitions are defined on startup by splitting alphabeth A-Z to a configured number of groups.
Wire protocol between client & server is a simple line protocol:
request: [A-Z]\n
response: [0|1]\n
Request is requested symbol; 0 response means rate limit is not breached, 1 -- rate limit is breached.
Other payloads considered invalid, however server returns 1 if it receives malformed message.
Server is located in server and is implemented in Rust using async-std and governor libraries.
Server binary accepts just two positional arguments: NUMBER_OF_PARTITIONS and RATE_LIMIT.
Server understands two env variables:
SERVER_ADDRESS-- bind address for a server (default:127.0.0.1:31337).SERVER_SEED-- seed for random generator.
To run server go to server and use cargo run [ARGS...] or build docker image using provided Dockerfile.
Client is located in clients/client.py and is implemented in Python (> 3.6). Only std-lib is used.
To run use python clients.py or execute it with ./clients.py.
argparse is used to implemented subcommands, so help will be provided.
Number of different behaviours are implemented, some of it are:
cycle-- cycles through available symbols with configured delay.scenario-- parses string argument as a sequence of messages and delays and executes that scenario.find_partitions-- runs reverse engineering algorithm, trying to figure out partitioning used by server.
Full list of supported behaviours is shown in program usage.
reverse-engineering/prototype.py contains network-less implementation of reverse engineering algo used in find_partitions client behaviour above.
docker-compose up builds server and client and runs find_partitions behaviours.