/PER2023-045

Deployment, monitoring, and performance optimization of a microservices technology stack on Kafka, Kubernetes, GCP.

Primary LanguageShell

PER2023–045 - Type : Developpement

Deployment, monitoring, and performance optimization of a microservices technology stack on Kafka, Kubernetes, GCP.

Déploiement, monitoring et optimisation de performances d'une pile technologique de Micro services, sur Kafka, Kubernetes, GCP

Authors of this repository

Technologies

  • Docker
  • Kubernetes (MiniKube, K3s)
  • Kafka
  • Ansible
  • Prometheus
  • Grafana
  • Python
  • Helm

Project structure

  • ansible folder which contains the .yaml files to automatically deploy the application and its environment
  • kubernetes folder which contains the .yaml files used by the Ansible scripts to deploy all the Kubernetes ressources
  • scripts folder which contains the following .sh files:
    • chmodAll.sh: give the execution rights to all the scripts
    • deploy-k3s-cluster.sh: deploy a K3s cluster on a list of hosts
    • deployEnv.sh: deploy all the environment needed for the application
    • launchExperience.sh: launch the execution of the application, retrieve the data and generate the graphs in the python/output folder
    • mnk-requirements1.sh: install Docker for Minikube
    • mnk-requirements2.sh: install Kubectl, Minikube, Helm, Python packages and Ansible

First configuration with Minikube on a single, clean machine

NB: You must be in the root folder to use the scripts.

Requirements

  • 18 CPU cores
  • 18 GB of RAM

Steps

  • Execute the following command in order to be able to execute all the scripts
chmod +x scripts/chmodAll.sh && scripts/chmodAll.sh
  • Install Docker
scripts/mnk-requirements1.sh

  • Once the previous script is done, exit the machine and reconnect to it in order to update the Docker users group, thus avoiding using the sudo su command everytime

  • Install Kubectl, Minikube, Helm, Python packages and Ansible

scripts/mnk-requirements2.sh
  • Deploy all the ressources
scripts/deployEnv.sh
  • Launch the experience
scripts/launchExperience.sh

Second configuration with K3s on a cluster of clean machines (Grid5000)

Requirements

  • An account on Grid5000

Steps

  • All information about Grid5000 can be found on getting started
  • Clone the project on your home directory on a site of Grid5000 (for exemple sophia)
  • Get the number of hosts that you want for the K8s cluster
# Exemple of 2 nodes for 2 hours
oarsub -I -l host=2,walltime=2 -t deploy
kadeploy3 debian11-min
  • Clone this project in your Grid5000 home directory
git clone https://github.com/antoinebqt/PER2023-045.git
  • Modify the IP addresses in K3s/hosts.ini (only one master is allowed)
  • Deploy the K3s cluster

NB: You must be in the root folder of the project to use the scripts.

cd PER2023-045
chmod +x scripts/chmodAll.sh && scripts/chmodAll.sh
vim k3s/hosts.ini
pip install ansible # not necessary if already installed
scripts/deploy-k3s-cluster.sh
  • Connect to your master node
ssh root@<ip_address>
ssh root@<grid_node_name>
  • Deploy the stack
cd ~/PER2023-045
scripts/deployEnv.sh
  • Wait 10 minutes then launch the experience
scripts/launchExperience.sh
  • Wait until the end of the experience
  • Retrieve the experience data
# from your home Grid5000, not the master node
scp -r root@<ip_address>:~/PER2023-045/python/output ~

The data will be in your home directory