The percentage of inject abort faults is not working as expected when fault.abort.percentage.value is less than 100.

The error occurs in a Guided Exercise of the the Building Resilient Microservices with Istio and Red Hat OpenShift Service Mesh course.

This is the link to the exercise: https://rol.redhat.com/rol/app/courses/do328ea-1.1.0/pages/ch06s02

The exercise is performed by students in a dedicated virtual classroom that includes an OCP cluster. These are the installed versions in the classroom:

• OCP 4.4.6
• OSSM 1.1.2
• Kubernetes v1.17.1+f63db30

The exercise deploys 3 services in OSSM: customer, preference, and recommendation.

The traffic flow is customer => preference => recommendation.

See app-deployment.yml  for more details.

Then, abort fault is configured on the recommendation service with a percentage of 50, using vs-recommendation-error.yml

After applying this configuration, waiting for 10 seconds, and sending 100 requests, no requests fail, although we would expect roughly 50% of requests to be aborted.

If we set the percentage fault.abort.percentage.value: 100, all requests fail.

If we set the percentage fault.abort.percentage.value: 90, around 30 out of 100 fail, which is far from the expected percentage