Skip to content

Latest commit

 

History

History
89 lines (77 loc) · 2.83 KB

README.md

File metadata and controls

89 lines (77 loc) · 2.83 KB
Raw

GoDFS

Very Simple Distributed FileSystem implementation on the lines of GFS and HDFS

NameNode

The NameNode is the master and is the single point of contact (and failure!) for coordinating all operations. It provides the clients with all the required metadata for accessing the DataNodes.

DataNode

The DataNode is a simple data containing unit and clients talk to them directly to put / get data into / from them.

Client

The short-lived client does all the talking to get operations done. A read / write client can be initiated through the command line interface provided.

Usage

One NameNode and at least one DataNode(s) must be initiated as daemons through the command line interface provided.

Run unit tests as:

make test

The CLI can be compiled to a binary to obtain godfs as:

make build

Natively

  • DataNode daemon Syntax:

     ./godfs datanode --port <portNumber> --data-location <dataLocation>

    Sample command:

     ./godfs datanode --port 7002 --data-location .dndata3/

  • NameNode daemon Syntax:

     ./godfs namenode --port <portNumber> --datanodes <dnEndpoints> --block-size <blockSize> --replication-factor <replicationFactor>

    Sample command:

     ./godfs namenode --datanodes localhost:7000,localhost:7001,localhost:7002 --block-size 10 --replication-factor 2

  • Client Currently Put and Get operations are supported

    • Put operation Syntax: 
       ./godfs client --namenode <nnEndpoint> --operation put --source-path <locationToFile> --filename <fileName>
      Sample command: 
       ./godfs client --namenode localhost:9000 --operation put --source-path ./ --filename foo.bar
    • Get operation Syntax: 
       ./godfs client --namenode <nnEndpoint> --operation get --filename <fileName>
      Sample command: 
       ./godfs client --namenode localhost:9000 --operation get --filename foo.bar

Containerized through Docker Compose

  • Build the images for the components: 
     docker build -t datanode -f daemon/datanode/Dockerfile .
 docker build -t namenode -f daemon/namenode/Dockerfile .
 docker build -t client -f daemon/client/Dockerfile .
  • Initiate the DataNode and the NameNode services (scale up accordingly): 
    docker-compose up --scale datanode=6 --remove-orphans --force-recreate
  • Start the client in a new container under the base host: 
    docker run -it --network host client
  • Make file put and get requests using similar commands as above

Todo

  • NameNode's regular HeartBeat check for DataNodes
  • Re-replication in case of dying DataNode
  • Secondary NameNode