Add automatic problem creation for scheduler failures

[nlr-ai]

Summary

Completes the self-healing infrastructure loop by automatically creating problem records when scheduled tasks fail, enabling Arsenale to detect and fix these issues autonomously.

What's New

Automatic Problem Creation

When any scheduled task fails, the scheduler now:

1. Sends a Telegram notification (existing behavior)
2. Creates a problem record in the PROBLEMS table (new)

Problem Details Include:

• Unique Problem ID: scheduler_failure_YYYYMMDD_HHMMSS_task_name
• Type: scheduler_task_failure
• Severity: High (scheduler failures are critical)
• Description: Script path, error message, and last 1000 chars of log output
• Suggested Solutions: Common fixes like checking paths, dependencies, env vars
• Assignment: ConsiglioDeiDieci (system problems)

Error Types Handled:

• ✅ Non-zero return codes
• ✅ FileNotFoundError (missing scripts)
• ✅ General exceptions
• ✅ Both 5-minute and hourly tasks

Duplicate Prevention

• Checks for similar problems created in the last hour
• Prevents spam when a task repeatedly fails

Integration with Arsenale

With this change, Arsenale can now:

1. OBSERVE: Detect scheduler failures in the PROBLEMS table
2. ASSESS: Analyze the error and design a fix
3. EXECUTE: Implement the solution (fix scripts, add missing files, etc.)
4. DOCUMENT: Verify the scheduler is working again

This creates a complete feedback loop where the infrastructure can heal itself!

Example Problem Record

{
  "ProblemId": "scheduler_failure_20250627_0315_citizen_activity_creation",
  "Type": "scheduler_task_failure",
  "Title": "Scheduler Task Failed: Citizen activity creation",
  "Description": "The scheduled task 'Citizen activity creation' failed...\n\nScript: /path/to/engine/createActivities.py\n\nError: Exception: No module named 'pyairtable'",
  "Status": "active",
  "Severity": "High",
  "Solutions": ["Check if the script exists...", "Review the error message...", "Check for missing dependencies..."]
}

---

This enhancement enables La Serenissima's infrastructure to autonomously detect and repair its own scheduled task failures, demonstrating true self-healing capabilities.

🤖 Generated with Claude Code