• A running OpenShift cluster is needed.
• The user should be authenticated to this cluster e.g., oc login
• Pre-requirements mentioned here should be met.

Get the IP address for the DNS resolver: oc describe dns.operator/default. Then set it in files such as:

• <path-of-repo>/socialNetwork/openshift/nginx-web-server-config/nginx.conf
• <path-of-repo>/socialNetwork/openshift/media-frontend-config/nginx.conf

Run the script <path-of-repo>/socialNetwork/openshift/scripts/deploy-all-services-and-configurations.sh

After customization, If you are running "on-cluster" copy necessary files to ubuntu-client, and then log into ubuntu-client to continue:

• ubuntuclient=$(oc -n social-network get pod | grep ubuntu-client- | cut -f 1 -d " ")
• oc cp <path-of-repo> social-network/"${ubuntuclient}":/root
  • e.g., oc cp /root/DeathStarBench social-network/"${ubuntuclient}":/root
• oc rsh deployment/ubuntu-client

• If using an off-cluster client:
  • Use oc -n social-network get svc nginx-thrift to get the cluster-ip.
  • Paste the cluster ip at <path-of-repo>/socialNetwork/scripts/init_social_graph.py:72
• If using an on-cluster client:
  • Use nginx-thrift.social-network.svc.cluster.local as cluster-ip and paste it at <path-of-repo>/socialNetwork/scripts/init_social_graph.py:72
• Register users and construct social graph by running cd <path-of-repo>/socialNetwork && python3 scripts/init_social_graph.py. This will initialize a social graph based on Reed98 Facebook Networks, with 962 users and 18.8K social graph edges.

There are three type of applications in the workload: compose post, write home timeline, and read home timeline. The applications run independently and each one has a different workflow, using different micro-services.

First, make sure that the wrk command has been properly built for your platform:

cd <path-of-repo>/socialNetwork/wrk2
make clean
make

For all load generating commands below, it should be possible to use nginx-thrift.social-network.svc.cluster.local:8080 for cluster-ip.

cd <path-of-repo>/socialNetwork/wrk2
./wrk -D exp -t <num-threads> -c <num-conns> -d <duration> -L -s ./scripts/social-network/compose-post.lua http://<cluster-ip>/wrk2-api/post/compose -R <reqs-per-sec>

cd <path-of-repo>/socialNetwork/wrk2
./wrk -D exp -t <num-threads> -c <num-conns> -d <duration> -L -s ./scripts/social-network/read-home-timeline.lua http://<cluster-ip>/wrk2-api/home-timeline/read -R <reqs-per-sec>

cd <path-of-repo>/socialNetwork/wrk2
./wrk -D exp -t <num-threads> -c <num-conns> -d <duration> -L -s ./scripts/social-network/read-user-timeline.lua http://<cluster-ip>/wrk2-api/user-timeline/read -R <reqs-per-sec>

Use oc -n social-network get svc jaeger-out to get the NodePort of jaeger service.

View Jaeger traces by accessing http://<node-ip>:<NodePort>

Original containers are expected to run as root, but OpenShift does not permit it as a default policy. As a workaround for simplicity, we can relax security policy by adding several SCCs to each project.

$ oc adm policy add-scc-to-user anyuid -z default
$ oc adm policy add-scc-to-user privileged -z default

If local image customization is needed, then the script <path-of-repo>/socialNetwork/openshift/scripts/build-docker-img.sh can be used to create them. In this case the relevant yaml files will need to be edited to refer to the new images. e.g., image: image-registry.openshift-image-registry.svc:5000/social-network/social-network-microservices:openshift