Too Long; Didn't Read
Ruby on Rails was facing a major memory issue with its production Ruby on rails. The system was a monolithic GraphQL API designed to handle routine business logic and consume various cloud services. Even during lulls in user traffic, our system's memory consumption continued rising, reaching levels equating to several GB. We tried a few "quick fixes" in our dev environment, including automatic reboots and automatic re-deployments, but this too was deemed impractical for use in production. No single block of code seemed to be responsible for driving memory up so high, but we decided to dig a little deeper.