Interested in approach to test random scenarios in case of E2E testing.
Q1: So we need to check all the system parts are connected correctly and what does it mean in case of random answer from the server etc (happy parts instead of single happy part)?
Q2: How to test errors in e2e scenarios? As example of this case different server errors etc. Does it have to be tested at all?
My experience and
approach to test random scenarios in case of E2E testing
is related to Gamifcation logic, which is hard to test, especially if you have to bend automation to do it. Randomness on every step is not the best case scenario for checking. I had similar issues while automating an questionnaire feature on a web platform and betting games on another.
Just to give you context - every challenge was loaded based on user level, most questions were influenced or did influence others.
After a lot of discussions and tryouts, it was clear that we should cover the main business (money) paths and leave the interesting part for exploratory testing. So, extract the most stable/predictable journeys and cover those with automation that is (reasonbly) insensitive to expected events and can record-retry steps in order to complete the scenario. My takeaway was, to find the right implementation-costs balance.