NetCommander

NetCommander is a client-server program developed in C language as part of the DV1457/DV1578 Lab 3 for the 2022 fall semester of lp1. The project focuses on implementing a server capable of handling multiple client connections concurrently. It supports two distinct operations: matrix inversion and k-means clustering. You can also run any command in the server from the client but the code needs some modifications in it. I have my code designed to mainly run the executable files so it always adds "./" to the commands sent to the server by the client and then executes it.

Project Structure

The project structure adheres to the following directory layout:

├── Client1
│   ├── client.c
│   ├── kmeans-data.txt
│   └── results
├── LICENSE
├── makefile
├── mathserver
│   ├── computed_results
│   ├── include
│   │   ├── kmeans.h
│   │   └── matrix_inverse.h
│   ├── objects
│   └── src
│       ├── kmeans.c
│       ├── kmeans-viz.py
│       ├── matrix_inverse.c
│       └── server.c
└── README.md

• The mathserver folder contains server-related source files.
• Header files are placed in the include folder, source code files in the src folder, and object files in the objects folder.
• The client folder contains all client-related files.
• The computed_results directory stores results generated by the server.
• The results folder holds individual client result directories.

Building and Running

To build the project, execute the following commands in the project root directory:

make

This will compile the server and client programs, placing the final executables in the project folder.

Running the Server

To start the server on a specified port (e.g., 9999), use the following command: First change your directory to mathserver/src and run the server code. I am using mainly fork because I have tested this more, I also have muxbasic and muxscale code in place but it needs some testing.

./server -p 9999 fork

The server will listen for incoming client connections.

Running a Client

To connect a client to the server, use the following command: Open a different terminal that is different from the server terminal and then change the directory to client1.

./client -ip <server_ip> -p <server_port>

Once connected, the client can send commands and receive results from the server.

Command Line Options

The server supports the following command-line options:

• -h: Print help text
• -p port: Listen on the specified port number
• -d: Run as a daemon
• -s strategy: Specify the request handling strategy (fork, muxbasic, muxscale)

The matinv-par.c and kmeans-par.c files must support all the command-line options supported by the sequential code.

Concurrent Client Handling

The server efficiently handles multiple client sessions concurrently using strategies like fork, muxbasic, or muxscale.

Implementation of Mathematical Operations

Matrix Inversion

The server calculates the inverse of a matrix using Gauss-Jordan elimination. The provided matrix_inverse.c program can be compiled and executed as follows:

gcc -w -O2 -o matinv matrix_inverse.c
./matinv -n 4 -P 1 -I fast

The server supports matrix inversion using the Gauss-Jordan elimination method. The client can send a matrix inversion command to the server, and the server computes and returns the inverted matrix. Matrix inversion is optimised using threads and you can initialize a matrix with nearly 2048x2048 size without in few seconds and then send the file as well without any transfer issues.

matinvpar -n 8 -I fast -P 1

Command:

matinvpar -n <size> -I <init_type> -P <num_threads>

Flags:

• -n <size>: Specifies the size of the matrix to be inverted. Example: -n 8 for a size 8x8 matrix.

• -I <init_type>: Specifies the initialization method for the matrix. Possible values for <init_type> include:

  • fast: Fast initialization.
  • rand: Random initialization.

• -P <num_threads>: Specifies the number of threads to be used for parallelization during matrix inversion.

Example:

matinvpar -n 8 -I fast -P 1

This example command requests the inversion of an 8x8 matrix using fast initialization with a single thread.

k-means Clustering

The k-means clustering algorithm is implemented to group data points into specified clusters. The provided kmeans.c program can be compiled and executed as follows:

gcc -w -O2 -o kmeans kmeans.c
./kmeans
python3 kmeans-viz.py

The server implements the k-means clustering algorithm for grouping data points into specified clusters. The client sends a k-means command with the data file, and the server computes and returns the cluster assignments.

kmeanspar -f kmeans-data.txt -k 9

Command:

kmeanspar -f <data_file> -k <num_clusters>

Flags:

• -f <data_file>: Specifies the path to the file containing the problem data for k-means clustering. Example: -f kmeans-data.txt.

• -k <num_clusters>: Specifies the number of clusters for k-means clustering.

Example:

kmeanspar -f kmeans-data.txt -k 9

This example command performs k-means clustering using the data in the file kmeans-data.txt with 9 clusters.

Contributing

Contributions to the project are welcome. Feel free to submit bug reports, feature requests, or pull requests to enhance the functionality.