Advisory Document Parsing Engine Heap Profiler and Memory Leak Isolation Script

Role: Lead Systems Performance Architect and Runtime Debugging Specialist focusing on large-scale advisory document extraction microservices.

Context

• Professional Services Practice: {{advisory_practice_unit}}
• Execution Runtime & Framework: {{runtime_environment}}
• OOM Crash Dump Summary: {{oom_crash_dump_summary}}
• Target Ingestion Throughput: {{target_throughput_tps}}
• Worker Concurrency Limit: {{worker_concurrency_limit}}
• Container Memory Ceiling: {{memory_ceiling_mb}}

Task

Design an end-to-end debugging and runtime heap profiling script for {{runtime_environment}} that isolates memory leaks during massive document parsing runs in {{advisory_practice_unit}}, triggers automated heap snapshots before {{memory_ceiling_mb}} exhaustion, and enforces graceful worker recycling.

Method

1. Analyze {{oom_crash_dump_summary}} to identify uncollected retained objects, closure leaks, and buffer allocation bottlenecks.
2. Construct an active runtime memory observer that tracks RSS, heapUsed, and external buffer allocations on a 250ms cadence.
3. Integrate trigger logic to generate V8/Python heap profiles automatically when memory consumption hits 85% of {{memory_ceiling_mb}}.
4. Build an asynchronous stream pipeline interceptor that enforces backpressure whenever {{target_throughput_tps}} overwhelms parser GC cycles.
5. Implement a supervisor worker management script that gracefully drains in-flight client document chunks before cycling saturated worker processes.
6. Generate a heap diff analysis routine to pinpoint lingering object retainers across consecutive document conversion batches.
7. Provide automated load-generation commands to simulate peak concurrency up to {{worker_concurrency_limit}}.

Constraints

• Profiling hooks MUST NOT add more than 3% CPU overhead during normal steady-state document parsing.
• The supervisor script MUST ensure zero document chunk drops during graceful worker recycling.
• Do not use non-standard native compilation dependencies unless pre-bundled in {{runtime_environment}}.
• Snapshots and logs MUST NOT write unencrypted client document text to shared disk partitions.

Output format

1. Leak Diagnostic Assessment: Concise breakdown of the retention tree causing memory bloat.
2. Production Profiler Script (heap_monitor.js or heap_monitor.py): Full script with event emitters, threshold listeners, and snapshot dumps.
3. Cluster Supervisor Configuration (cluster_manager.js or supervisor.py): Zero-downtime worker lifecycle script.
4. Post-Mortem Verification Guide: 4-step checklist to validate memory stability under load.

Self-review

• Is backpressure correctly handled during document buffer transformation?
• Does the worker recycle routine drain in-flight promises/tasks before terminating the process?
• Are memory snapshots securely written to restricted temporary paths?