prerito / deploy-dc

Deploy Distributed Containers via k8s client-go Libraries + Custom Controller

Steps to Deploy DeathStar using Go Script

This readme is meant to provide insight into setting up and running the media-microservices DeathStarBench on a Cloudlab instance with 3 nodes (1 master, 2 worker nodes)

It has also been tested with the social-network micrroservices application. The only change is in the namespace that you have to create (i.e. create namespace social-network)

Setup Steps:

1. Install Docker and Kubernetes and Go
2. Instantiate a K8s cluster
3. Install python3-pip, asyncio, and aiohttp
4. Install k8s client go libraries:

       go get k8s.io/api
       go get k8s.io/apimachinery
       go get k8s.io/client-go
   

Deploy Application without Go Script

1. Change lines 55, 59, 63, and 67 in <path-to-repo>/DeathStarBench/mediaMicroservices/k8s-yaml/nginx-web-server.yaml which refers to the the installation directory location of DeathStarBench to the appropriate location
2. On Master note, create namespace media-microsvc via: kubectl create namespace media-microsvc
3. Deploy all pods via: kubectl apply -f <path-of-repo>/mediaMicroservices/k8s-yaml/
4. Wait until kubectl -n media-microsvc get pod displays all Pods as running

Deploy Application with Go Script

1. Change lines 55, 59, 63, and 67 in <path-to-repo>/DeathStarBench/mediaMicroservices/k8s-yaml/nginx-web-server.yaml which refers to the the installation directory location of DeathStarBench to the appropriate location
2. On Master note, create namespace media-microsvc via: kubectl create namespace media-microsvc
3. Deploy all pods via: go run main.go -f <path-to-app-def-json>
   • assure that the app-def JSON file has the correct filepath to the deployment application
4. Wait until kubectl -n media-microsvc get pod displays all Pods as running

Register users and movie information

1. Use kubectl -n media-microsvc get svc nginx-web-server to get its cluster-ip.
2. Paste the cluster ip at <path-of-repo>/mediaMicroservices/scripts/write_movie_info.py:99 and <path-of-repo>/mediaMicroservices/scripts/register_users.sh:5 3.Update <path-of-repo>/mediaMicroservices/scripts/write_movie_info.py:95 & 97 to point to the installation path of the repo
3. python3 <path-of-repo>/mediaMicroservices/scripts/write_movie_info.py && <path-of-repo>/mediaMicroservices/scripts/register_users.sh

Running HTTP workload generator

A sample workload can be to compose reviews - to do this, do:

1. Paste the cluster ip at <path-of-repo>/mediaMicroservices/wrk2/scripts/media-microservices/compose-review.lua:1032

2. cd <path-of-repo>/mediaMicroservices/wrk2
   ./wrk -D exp -t <num-threads> -c <num-conns> -d <duration> -L -s ./scripts/media-microservices/compose-review.lua http://<cluster-ip>:8080/wrk2-api/review/compose -R <reqs-per-sec>