PID Memory Inspector

PID Memory Inspector is a powerful visualization and diagnostics tool for tracking memory usage per process over time — with full support for process trees and snapshots.

---

✨ Features

• 📈 Live-like graphs of RAM and Swap usage
• 🌲 Tree-based process snapshots at any point in time
• 🔍 Drill-down into any PID: its own memory + all children
• 📊 Graphs: own RSS, subtree RSS, children trends
• 🧠 Filter by lifetime, RSS thresholds, subtree memory
• 🧭 Per-level tree exploration (level 0 / 1 / 2 / ...)
• 🔗 Clickable PID / PPID navigation
• 📎 Auto-aggregated summaries and stats
• 📁 Works with CSV dumps from dumps/time/

---

🚀 How it works

1. Collect memory usage dumps with:

   python scripts/collect_memory.py

   This periodically creates:

   • sys_mem_*.csv — system metrics (/proc/meminfo)
   • process_mem_*.csv — processes (ps dump)

2. Run the app:

   make run
   # or
   uvicorn src.app:create_app --reload --factory

3. Open in browser: http://localhost:8000

---

🧩 Interface overview

• /api/v1/ — Home dashboard: RAM + Swap graphs
• /api/v1/snapshot/level?lvl=N — Processes at level N
• /api/v1/snapshot/pid?pid=... — Explore subtree of a given PID
• /api/v1/snapshot/pid/plot?pid=... — Graphs + stats for PID + children

All views are interactive, filterable, and linked via PID navigation.

---

🔧 Project structure

• scripts/collect_memory.py — CSV memory dumper
• src/application/ — orchestration layer (MetricsService)
• src/domain/ — core logic: tree stats, filters, timelines
• src/interfaces/web/ — FastAPI routes, HTML + Plotly
• static/mem.css — table styling

---

🤖 Tech stack

• Python 3.11 + FastAPI
• Pandas for data handling
• Plotly for visualizations
• PureCSS for lightweight styling
• Custom layered architecture

---

🗂️ Example filters

• lifetime ≥ 300s
• RSS ≥ 100 MB
• subtree_rss ≥ 500 MB

---

🔮 Future ideas

• Per-process Swap: from /proc/<pid>/smaps_rollup
• CGroup-aware memory rollups
• Export to JSON / PNG / CSV
• Alerting on anomalies (spikes, leaks)

---

Made with ♥ for anyone who's ever had to answer:

"Wait... what process is eating all the RAM?"