System Analysis Report
Documentation for Developers
System Performance Analysis Report
Date: Feb 10, 2026
Uptime: 3 days, 4 hours
Machine: 24-core ARM Mac, 128GB RAM
Critical findings
1. Load average: 39.53 (1-minute)
- 24 cores; load ~39 ≈ 1.6× overload
- Target: under 24
- Indicates many processes competing for CPU
2. Cursor main renderer: ~13 GB RAM, 40–60% CPU
- PID 6501 — main Cursor window
- High memory and CPU suggest:
- Large workspace or many open files
- Possible memory leak (long uptime)
- Heavy extensions (Angular, ESLint, MongoDB, etc.)
3. Chrome: 86 processes, ~13 GB RAM
- Main process: ~31% CPU, 5.7% memory
- Many renderer processes (one per tab/extension)
- Heavy GPU process (~18% CPU)
4. Duplicate MCP servers
- 4+ gspace1 MCP server instances (started 9:16 PM and 10:28 AM)
- 3 editor2 MCP servers (for Claude)
- Extra instances add CPU and memory
5. Multiple Cursor renderer windows: 5
- Each renderer = its own memory and CPU
- Implies multiple workspaces/windows (gspace1, gmodules1, etc.)
6. Heavy extensions (43 Cursor-related processes)
- Angular Language Service (gspace1 + gmodules1)
- ESLint, MongoDB, JSON, HTML, Markdown
- Docker/Containers, Auto Rename Tag
- retrieval-always-local, agent-exec
- Nx daemon, nx-mcp, playwright-mcp
7. Three backends running
- backend-api, backend-ai-api, backend-vitto-api
- Normal for dev, but adds ~8% CPU combined
8. Other resource usage
- Spotlight (mds_stores): ~3.8 GB — indexing can spike CPU during changes
- Virtualization.VirtualMachine: ~4.3 GB — VM or container
- ~70,000 open file descriptors — many watchers/connections
- 1,100 total processes
What’s healthy
- Memory pressure: ~93% free
- No swap activity (swapins/swapouts: 0)
- Nx cache: 43 MB
- No runaway Node processes (after previous cleanup)
Recommended actions (in priority order)
| Priority | Action | Impact |
|---|---|---|
| 1 | Restart Cursor | Frees ~13 GB, may fix renderer leak; expect ~5 min downtime |
| 2 | Close Chrome tabs | Aim for <20; use tab suspender or bookmarks for rarely used tabs |
| 3 | Run fewer backends | Use only backend-api + backend-ai-api when needed; stop backend-vitto-api if unused |
| 4 | Close extra Cursor windows | One main workspace reduces processes and load |
| 5 | Exclude node_modules from Spotlight |
Reduce indexing in large monorepos |
| 6 | Disable or limit heavy extensions | Especially when not needed (e.g., MongoDB, Docker, extra language servers) |
| 7 | Reboot periodically | Uptime 3+ days can accumulate leaks and stale processes |
Spotlight exclusions (optional)
For large dev folders:
# Add node_modules and dist to Spotlight exclusions
sudo mdutil -i off /Users/fguijarro/developer/repos-gspace/gspace1/node_modules
sudo mdutil -i off /Users/fguijarro/developer/repos-gspace/gspace1/dist
# Re-enable if needed: sudo mdutil -i on /path
Generated by system analysis. Run nx reset && rm -rf dist/ tmp/ if build issues persist.