Redis Clone

This is my solution to the "Build Your Own Redis" Challenge.

Running the redis server

Spawning a new server instance (--port is optional, defaults to port 6379):

$ ./spawn_redis_server.sh --port 6379

Spawning a new server instance, loading a saved rdb file data located at /tmp/redis/dump.rdb:

$ ./spawn_redis_server.sh --dir /tmp/redis-files --dbfilename dump.rdb

Spawning a new replica of an existing server @localhost:6379:

$ ./spawn_redis_server.sh --port 8899 --replicaof localhost 6379

Features

PING, ECHO

The server can respond to PING and ECHO command.

On the first terminal:

$ ./spawn_redis_server.sh

On the second terminal:

$ redis-cli ping                  # "PONG"
$ redis-cli echo "Hello world!"   # "Hello world!"

SET, GET

The server can respond to SET and GET command.

On the first terminal:

$ ./spawn_redis_server.sh

On the second terminal:

$ redis-cli set foo bar   # "OK"
$ redis-cli get foo       # "bar"

The server also supports SET with timeout:

// Setting timeout to 5000 ms
$ redis-cli set foo bar px 5000   # "OK"
$ redis-cli get foo               # "bar"
// after 5 seconds
$ redis-cli get foo               # (nil)

Replication

All data are forwarded to each replica. A client can send a SET command to the master and the data will be duplicated on the other replicas.

On the first terminal:

$ ./spawn_redis_server.sh
$ ./spawn_redis_server.sh --port 8899 --replicaof localhost 6379

On the second terminal:

$ redis-cli set foo bar
// talk to the replica instead
$ redis-cli -p 8899 get foo     # "bar"

The server also supports the WAIT command. WAIT N TIMEOUT will wait for TIMEOUT milliseconds or until the server receives an acknowledgement from at least N servers, whichever is faster.

Continuing from previous scenario, on the second terminal:

$ redis-cli set key value
$ redis-cli WAIT 1 500  # returns 1 immediately
$ redis-cli WAIT 3 1000 # returns 1 after timeout of 1 second

Persistence: Reading from a .rdb file

The server can read a .rdb file generated by a redis server (https://redis.io/download/) and load its content.

Generating a .rdb file:

On the first terminal:

$ redis-server

On the second terminal:

$ redis-cli set foo bar
$ redis-cli save

Then, press Ctrl+C on the first terminal. A file dump.rdb will be created.

The server can use this .rdb file to update its key-value store:

On the first terminal:

$ ./spawn_redis_server.sh --dir . --dbfilename dump.rdb

On the second terminal:

$ redis-cli get foo       # "bar"

Persistence: Writing to a .rdb file

The server can also persist its key-value store to a .rdb file which can be read by a redis server.

Client needs to send the "save" command to instruct the server to write its key-value store to disk upon exit.

On the first terminal:

$ ./spawn_redis_server.sh

On the second terminal:

$ redis-cli set foo bar
$ redis-cli save

Then, press Ctrl+C on the first terminal. A file TIMESTAMPdump.rdb will be created. TIMESTAMP will be in YYMMDD_HHMMSS format.

Simply rename this file to dump.rdb and a redis server will be able to load the contents.

$ redis-server