Debugging a Python "Memory Leak" That Was Actually a Measurement Bug (ru_maxrss vs VmRSS)

The author encountered a perplexing memory leak in a Python script used to build a RAG knowledge base, suspecting the embedding loop was responsible. Despite trying various memory management techniques, the issue persisted. Eventually, it was discovered that the memory usage metrics were misleading due to a misunderstanding between ru_maxrss and VmRSS, revealing a measurement bug rather than an actual leak. This highlights the importance of careful metric interpretation in debugging and the potential for costly misdiagnoses.