Partial implementation of the Raft consensus algorithm1.
persistence, snapshot
make start_clustermake stop_clusterAn up to date node should always be available.
You can find the leader from the kv_client.py output.
You can use docker-compose stop $peer_name to kill a leader and force a switch over
with docker-compose start $peer_name you can start the peer again and let it join the cluster.
Add the pyraft package to your PYTHONPATH and execute the command from the project directory.
To read a key
env PYTHONPATH=./pyraft:$PYTHONPATH python ./pyraft/client/kv_client.py -nodes 127.0.0.1:7001,127.0.0.1:7002,127.0.0.1:7003 -get your_keyTo add a key
env PYTHONPATH=./pyraft:$PYTHONPATH python ./pyraft/client/kv_client.py -nodes 127.0.0.1:7001,127.0.0.1:7002,127.0.0.1:7003 -put your_key valueTo remove a key
env PYTHONPATH=./pyraft:$PYTHONPATH python ./pyraft/client/kv_client.py -nodes 127.0.0.1:7001,127.0.0.1:7002,127.0.0.1:7003 -del your_key # start the cluster, you can find the leader from the output
make start_cluster# add a new key, from a different terminal, you can find the leader id from the command output
env PYTHONPATH=./pyraft:$PYTHONPATH python ./pyraft/client/kv_client.py -nodes 127.0.0.1:7001,127.0.0.1:7002,127.0.0.1:7003 -put foo bar
# read the new added key
env PYTHONPATH=./pyraft:$PYTHONPATH python ./pyraft/client/kv_client.py -nodes 127.0.0.1:7001,127.0.0.1:7002,127.0.0.1:7003 -get foo
# kill the leader node
docker-compose stop $peer_name
# add a couple of keys and start the node again, verify your keys
docker-compose stop $peer_name