Deadlock project for MEC

This is a property of London South Bank University licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)

Please get permission before use

@Author: Emeka Emmanuel Ugwuanyi ugwuanye@lsbu.ac.uk with contributions from Saptarshi Ghosh

Note for self

(((23**r.randrange(1, 1331)) % r.randrange(1, 1777)) % 5)

ssh setup

ssh-keygen -t rsa ; ssh-copy-id mec@192.168.122.110

SET UP Commands

apt update  
passwd   
password   
password   
apt install git sudo -y  
adduser mec  
password  
password  
usermod -aG sudo mec  
su - mec  
git clone https://github.com/emylincon/deadlock_project.git  
cd deadlock_project  
sudo bash set_up.sh  
password

How To Use The System

Run instructions for heterogeneous environment

• The following are the steps for running and setup of the system.
• The files for heterogenous are located in 2algo folder

GNS3 Heterogeneous Architecture

SDN Architecture

Broker:

• start the mosquitto server first.
• it is usually located on the first MEC node

/etc/init.d/mosquitto start

cloud node:

• start the cloud nodes.

python3 deadlock_project/2algo/0_cloud.py

• then enter the brokers ip (ip address on eth0)

MEC nodes:

navigate to the MEC user. that is where the deadlock_project is sitting

cd /deadlock_project/2algo/

run any of the desired algorithm

• the speak .py runs only on the broker with is usually the first MEC
• the gui.py runs only on a gui not a server
• the NS.py runs on any other node

Live Result During Runtime

Clients:

• if the dns does not work for the client in ubuntu 18 run sudo bash client_dns.sh
• each client(o_gui_user.py) have a prepared data they have to read and run from {user3, user4 and user5}
• each correspond to a respective client node
• the corresponding python file have to be renamed to record.py in the user node.
• there are 3 official client/user files:
• 1.) 0.0_user.py : this is for the server
• 2.) gui_user.py : this is for the desktop. it shows a live plot of the results obtained. Run this file if you dont want to use a predefined data. it automtically generates data using GG1 distribution
• 3.) o_gui_user.py : This is also to be used on a desktop because it shows a live plot of the result obtained. the difference between this and gui_user is that the this is used when you already have a predefined data. the data should be named record.py and placed on each of the client node

Run instructions for homogeneous environment

• The following are the steps for running and setup of the system.
• The files for homogeneous are located in algo folder

cloud node:

• start the cloud nodes.
• run the setup.sh script to download necessary files for the algorithms to run

sudo bash setup.sh 

• first start the ssh server on the cloud node
• then run the python script.

python3 deadlock_project/algo/0_cloud.py

MEC nodes:

navigate to the MEC user. that is where the deadlock_project is sitting

cd /deadlock_project/algo/

• run the setup.sh script to download necessary files for the algorithms to run

sudo bash setup.sh 

• start ssh server
• run any of the desired algorithm
• the speak .py should be executed first
• the gui.py runs only on a gui not a server
• the NS.py runs on any other node
• all the files need a predefined distribution data to run except the files that have the letter 'n' just after the number
• the predefined data available in deadlock_project/algo/Experiment could be used
• the data available are labelled data.py, data1.py, data2.py and data4.py
• for any of these to be used, it has to be renamed to data.py and moved to the directory deadlock_project/algo

GNS3 Homogeneous Architecture

SDN Architecture

Algorithms implemented

• RMS + Bankers
• EDF + Bankers
• RMS + Wound Wait
• RMS + Wait Die
• EDF + Wound Wait
• EDF + Wait Die

Evolution of bankers algorithm