This module will run automated Chaos Tests for the SharedInformers

Start minikube, e.g.:

minikube start --driver=docker --memory 8192 --cpus 3

Install ChaosMesh on minikube:

curl -sSL https://mirrors.chaos-mesh.org/v2.6.0/install.sh | bash

Wait for the pods to be all ready:

kubectl wait --for=condition=Ready pods -n chaos-mesh --all --timeout=600s

Build the control and checker Docker images in the minikube docker-env and run the test:

eval $(minikube -p minikube docker-env)
mvn -pl chaos-tests -Pitests -Phttpclient-jdk verify

• checker: run the SharedInformer to get notifications over the shared resource
• control: apply timely changes to the shared resource

As of today this is the main flow of the tests(please double-check the implementation if this documentation is not accurate anymore):

• Starts a "Checker" application(as a Pod in a cluster) which starts a SharedInformer on a ConfigMap
• Starts a "Control" application(as a Pod in the cluster) which keeps incrementing a counter stored in the previously mentioned ConfigMap
• The "Checker" is waiting for the number to reach a specific value N
• The "Control" keeps incrementing the counter up to the same number N
• We start a Chaos Experiment targeting the "Checker" app, in this setup, we are using 3 different Network disruption scenarios
• The Chaos Experiment lasts for 12 minutes
• After the Experiment is finished we do expect that the "Checker" app SharedInformer recovers and eventually reaches the target N
• Eventually, both the 2 Control and Checker Pods should be in a Successful state

We are testing the resiliency of the WebSocket communication used under the cover by the SharedInformers and we guarantee that they do recover appropriately when something happens. We guarantee that we don't have regressions and this tests can be used as a gate for other HTTP client implementations.