system-restoration（系统恢复）

System Restoration

Comprehensive guide for restoring Advantage HPE's operational intelligence systems when they fail or go down.

Investigation Workflow

1. System Status Assessment

Before fixing anything, map out what's broken:

Core Intelligence Systems:

1. Zero Revenue Alerts → #margin-alerts (Every 30 min)
2. Morning Pulse → #manager-nudges (Daily 6:35 AM)
3. Live Nudges → #manager-nudges (Every 15 min)
4. Material Truth Report → #material-intel-systems (Daily 7:00 AM)
5. Friend-Zone Reformatter → #live-ops (ServiceTitan email alerts)

Investigation Commands:

# Check LaunchD services
launchctl list | grep ranger

# Check cron jobs  
cron list

# Check running processes
ps aux | grep -E "(keel|pulse|margin|nudge)" | grep -v grep

# Find system code
find /Users/stephendobbins/.config/ranger -name "*.py" | grep -E "(pulse|margin|nudge)"
find /Users/stephendobbins/.openclaw/workspace -name "*.py" | grep -E "(zero|revenue)"

2. Locate Code & Determine Failure Cause

Common Locations:

• /Users/stephendobbins/.config/ranger/scripts/ - Main operational scripts
• /Users/stephendobbins/.config/ranger/materials/ - Material intelligence
• /Users/stephendobbins/.openclaw/workspace/ - Recent scripts & fixes
• /Users/stephendobbins/Library/LaunchAgents/ - LaunchD service definitions

Common Failure Patterns:

• LaunchD services unloaded - Emergency shutdown or system restart
• Data source broken - ServiceTitan API returning wrong data
• Scheduling missing - Functions exist but no cron/LaunchD trigger
• Script errors - Import failures, credential issues

System-Specific Restoration

Zero Revenue Alerts

Script: /Users/stephendobbins/.config/ranger/scripts/margin_alerts.py Channel: #margin-alerts (C0A5L7MG60P) Schedule: Every 30 minutes

Restoration Steps:

1. Verify script exists and posts to Slack
2. Load LaunchD service: launchctl load /Users/stephendobbins/Library/LaunchAgents/com.ranger.margin-alerts.plist
3. Test manually: cd /Users/stephendobbins/.config/ranger/scripts && python3 margin_alerts.py
4. Check logs: tail /tmp/margin_alerts.log

Morning Pulse

Script: /Users/stephendobbins/.config/ranger/scripts/pulse_os_full.py Channel: #manager-nudges (C0A5V9JL2KV) Schedule: Daily 6:35 AM CT

Restoration Steps:

1. If broken API data: Check for .bak backup with working data sources
2. Restore backup: cp pulse_os_full.py.bak pulse_os_full.py
3. Fix data sources: Replace API calls with browser automation (see references/browser-data-sources.md)
4. Load LaunchD service: launchctl load /Users/stephendobbins/Library/LaunchAgents/com.ranger.morning-pulse.plist
5. Test: python3 pulse_os_full.py pulse

Live Nudges

Script: /Users/stephendobbins/.config/ranger/scripts/pulse_os_full.py nudges Channel: #manager-nudges Schedule: Every 15 minutes

Function: run_nudges() on line 548-617 Features: 🚗 dispatched / 📍 arrived / ✅ completed alerts

Restoration Steps:

1. Verify function exists: grep -n "def run_nudges" pulse_os_full.py
2. Create LaunchD service (see scripts/create-live-nudges-service.py)
3. Load service: launchctl load /Users/stephendobbins/Library/LaunchAgents/com.ranger.live-nudges.plist
4. Test: python3 pulse_os_full.py nudges

Material Truth Report

Script: /Users/stephendobbins/.config/ranger/materials/reconciliation_report.py Channel: #material-intel-systems (C0A5L7RB5EK) Schedule: Daily 7:00 AM CT

Restoration Steps:

1. Test script: cd /Users/stephendobbins/.config/ranger/materials && python3 reconciliation_report.py --no-email
2. Create cron job with 7:00 AM schedule
3. Verify channel posting

Data Source Repair

ServiceTitan API vs UI Data

Problem: ServiceTitan API often returns test/historical data instead of real operational data.

Solution: Replace API calls with browser automation:

1. Create browser data source module (see scripts/browser_data_sources.py)
2. Import in main script: Replace parse functions with browser equivalents
3. Preserve output format - Same sections, different data source

Browser Data Functions:

• get_browser_low_margin_jobs()
• get_browser_stale_estimates()
• get_browser_revenue_leaks()
• get_browser_driver_incidents()

KEEL System Issues

Script: /Users/stephendobbins/.config/ranger/keel/keel_slack_bot.py

Safe restart for field tech DM only:

1. Disable operational intelligence: Set OPERATIONAL_INTELLIGENCE_ENABLED = False
2. Restart process: cd /Users/stephendobbins/.config/ranger/keel && python3 keel_slack_bot.py &
3. Verify running: ps aux | grep keel_slack_bot

Service Management Commands

LaunchD Services

# List services
launchctl list | grep ranger

# Load service  
launchctl load /Users/stephendobbins/Library/LaunchAgents/com.ranger.<service>.plist

# Unload service
launchctl unload /Users/stephendobbins/Library/LaunchAgents/com.ranger.<service>.plist

# Start service immediately
launchctl start com.ranger.<service>

# Check service logs
tail /tmp/<service>.log
tail /tmp/<service>.err

Cron Jobs (OpenClaw)

# List jobs
cron list

# Add job  
cron add <job-definition>

# Remove job
cron remove <job-id>

Emergency Shutdown Recovery

When systems are emergency-stopped due to bad data:

1. Investigate root cause - Usually ServiceTitan API data issues
2. Fix data sources - Switch to browser automation or correct API endpoints
3. Test manually - Verify data accuracy before re-enabling
4. Restore services - Load LaunchD services and cron jobs
5. Monitor initially - Check logs and channel posts for accuracy

Resources

scripts/

• create-live-nudges-service.py - Generate LaunchD plist for live nudges
• browser_data_sources.py - Browser automation replacement for broken APIs

references/

• launchd-service-templates.md - LaunchD plist templates for different schedules
• channel-ids.md - Slack channel IDs for all operational intelligence channels
• troubleshooting-checklist.md - Step-by-step debugging guide