Enterprise Drupal / March 2026
The code was correct. The configuration was not the same.
In a classic enterprise failure, local and test environments behaved correctly while production-adjacent behaviour differed. The investigation eventually led to configuration drift.
Situation
One of the most expensive classes of bug is when the same commit behaves differently across environments. It is easy to keep rereading the code even though the difference actually lives in configuration, cache state, a feature flag or an environment variable. Several March support cases reinforced how important environmental diffing is.
Approach
I did not stop at the stack trace. I compared active configuration, deployment-sourced configuration, cache state and relevant environment values. Finding the difference was only half the task; I also traced how it could have appeared, such as a manual production change, a missed import or a different secret or configuration source.
Outcome
Individual incidents closed faster, and more importantly, several sources of drift became visible. The fix was not a manual production click but a change that the next deployment could reproduce as well.