Well, the recovery from my ghastly QA Epic Fail has been just as WEIRD as the original problem. First of all, when our BA mentioned “Criteria 1” he was talking about the main gist of the change — driver age was still getting populated with the age at the time of exposure creation and not the age at loss. I scoured the code in GitHub and the code my coworker added was just a bunch of loggers to troubleshoot HIS issue with his code, so that wouldn’t affect anything with my code.
My long-suffering coworker who’s been helping me out (let’s call him Mentor #2) had me bring the March Branch into my local (I KNEW this was going to happen), take the payload from our BA’s claim, and plop it into my JUnit test to see which age would appear.
To back this up a bit, the payloads created by our process, that are fed into the integration jobs, have changed format. I was able to escape using these last time by using an older claim for my JUnit tests—no such luck this time, as our BA had created this new claim just for his testing. I had several false starts where the damn JUnit test just didn’t work with the new payload format. Mentor #2 had to dial in for a meeting, so left me to my own devices, telling me to keep trying. After cursing and swearing that the damned payload was a bust, and that he had NO IDEA what he was talking about, and why on earth did they have to change the damned format, and…
…and then I saw it. Yours truly forgot to change the JUnit code to use the claim and policy for the BA’s claim—mine was still in there. Once I fixed this, the JUnit test ran like a Swiss watch…and wouldn’t you know, the correct age popped out! So, the issue wasn’t my code. Seeing that our BA originally created his claim February 4, I decided to take a chance that this might have been a system glitch—especially seeing as my coworker had had his code issue around the same time. Following the QA test scenarios, I created two more claims from the UI. This morning, I checked the final payload that goes to the next system, and again…I got the correct age.
SO, what happened? My tech lead believes it may be due to our BA running his tests before the test Spring Batch cycle ran, but my code was there on January 27 and worked just fine. I checked the output for the Spring Batch Build for Monday and the result said “UNSTABLE,” so perhaps something happened that effected our code.
For now, I’m going to lay low…but I think I’m going to need to keep running my tests every time this code hits a new test region and production.
