[Feature]: Metrics Enhancement (export all data, capture all metrics, fix last used timestamps, UI improvements)

[crivetimihai]

Metrics Enhancement Todo & Verification Checklist

Overview

This document tracks the remaining work for metrics enhancement after merging PR #657, which implemented the Metrics Tab UI enhancements as requested in issue #368.

Context

• Original Issue: #368 - Enhance Metrics Tab UI for MCP Context Forge
• Current PR: #657 - Metrics Tab UI Enhancements
• Branch: feature/metrics-tab-enhancements

Completed Work (PR #657)

✅ Added Virtual Servers to metrics display
✅ Created enhanced Top 5 Performance Tables with detailed columns
✅ Implemented tab navigation for different entity types
✅ Added responsive and mobile-friendly tables
✅ Fixed authentication dependencies and SQLAlchemy syntax
✅ Resolved Alembic migration conflicts
✅ Improved test coverage (1403 tests passing)

Outstanding Issues to Address

1. Export Functionality - Export ALL Data, Not Just Top 5

Current Behavior

• Location: mcpgateway/static/admin.js:1617-1662 (exportMetricsToCSV function)
• Issue: Export button only exports the top 5 items passed to it from topData
• Affected endpoints:
  • /admin/metrics page export button
  • /metrics API endpoint (if it exports)

Required Changes

• Modify export functionality to fetch ALL metrics data, not just top 5
• UI should continue showing top 5 for performance/clarity
• Export should include complete dataset

Implementation Details

// Current implementation in admin.js:1629-1650
["tools", "resources", "prompts", "gateways", "servers"].forEach((type) => {
    if (topData[type] && Array.isArray(topData[type])) {
        // Only exports what's in topData (limited to 5 items)
    }
});

Files to Check/Modify

• mcpgateway/static/admin.js - exportMetricsToCSV function
• mcpgateway/admin.py:4046-4049 - Currently limits to 5 in API response
• Service layer methods:
  • mcpgateway/services/tool_service.py:193 - get_top_tools(limit=5)
  • mcpgateway/services/resource_service.py:116 - get_top_resources(limit=5)
  • mcpgateway/services/prompt_service.py:143 - get_top_prompts(limit=5)
  • mcpgateway/services/server_service.py:131 - get_top_servers(limit=5)

2. Metrics Accuracy Verification

Areas to Verify

A. Last Used Timestamps

• Check: Ensure last_execution or lastExecution fields are properly updated
• Location:
  • Frontend: admin.js:1348-1372 (formatLastUsed function)
  • Backend: Metric model properties in mcpgateway/db.py
• Test: Execute tools/resources/prompts and verify timestamp updates

B. Success Rate Calculations

• Current Implementation: admin.js:1332-1342 (calculateSuccessRate function)
• Check for:
  • Division by zero protection (currently handled: total > 0 ? ... : 0)
  • Correct percentage calculation
  • Proper field mapping (successRate vs calculated value)
• Backend Sources:
  • Tool metrics: mcpgateway/db.py:527 (avg_response_time property)
  • Resource metrics: mcpgateway/db.py:745 (avg_response_time property)
  • Prompt metrics: mcpgateway/db.py:929 (avg_response_time property)
  • Server metrics: mcpgateway/db.py:1073 (avg_response_time property)

C. Execution Counts

• Verify: Increment logic for all entity types
• Check:
  • Tools: execution_count increments on tool invocation
  • Resources: execution_count increments on resource access
  • Prompts: execution_count increments on prompt usage
  • Servers: execution_count increments on server interaction

3. Response Time Truncation

Current Issue

• Response times show excessive decimal places: 1.1898999695965489
• Should be truncated to 3 decimal places: 1.189

Locations to Fix

• Frontend Display: admin.js:1642-1644

  item.avg_response_time || item.avgResponseTime
      ? `${Math.round(item.avg_response_time || item.avgResponseTime)}ms`
      : "N/A"

  Currently uses Math.round() which removes all decimals. Should use:

  `${(item.avg_response_time || item.avgResponseTime).toFixed(3)}ms`

• Backend Response: Check if truncation should happen at API level

  • Service methods returning avg_response_time
  • Schema definitions in mcpgateway/schemas.py

4. Performance & Edge Cases

Division by Zero

• Status: Already handled in frontend (total > 0 ? ... : 0)
• Verify: Backend calculations in metric properties

Large Dataset Performance

• Check: Performance with 1000+ metrics entries
• Consider: Pagination for export functionality
• Test: Load testing with concurrent metric updates

Database Query Optimization

• Review: Query performance in service layer get_top_* methods
• Check: Proper indexing on metric tables
• Verify: Efficient aggregation queries

Null/Undefined Handling

• Frontend: Check all field access for null safety
• Backend: Verify Optional fields handle None properly

Testing Checklist

Manual Testing

• Create multiple tool executions, verify metrics update
• Create resource accesses, verify metrics update
• Use prompts, verify metrics update
• Interact with servers, verify metrics update
• Check "Last Used" shows correct relative time
• Verify success rates calculate correctly (successful/total * 100)
• Export metrics and verify CSV contains ALL data, not just top 5
• Check response times are truncated to 3 decimal places
• Test with no metrics (empty state)
• Test with failed executions (success rate < 100%)

Automated Testing

• Unit tests for metric calculation methods
• Integration tests for metric endpoints
• Test export functionality with large datasets
• Test edge cases (division by zero, null values)

Performance Testing

• Load test with 1000+ metric entries
• Concurrent metric updates
• Export performance with large datasets
• UI responsiveness with many metrics

Implementation Priority

1. High Priority

   • Fix export to include ALL data (not just top 5)
   • Truncate response times to 3 decimal places
   • Verify success rate calculations

2. Medium Priority

   • Verify last used timestamps update correctly
   • Add comprehensive test coverage
   • Performance optimization for large datasets

3. Low Priority

   • Additional UI enhancements
   • Advanced filtering options
   • Real-time metric updates

Code Locations Reference

Frontend Files

• mcpgateway/static/admin.js - Main UI logic
• mcpgateway/templates/admin/metrics.html - Metrics page template

Backend Files

• mcpgateway/admin.py:4044-4051 - Metrics API endpoint
• mcpgateway/services/tool_service.py - Tool metrics logic
• mcpgateway/services/resource_service.py - Resource metrics logic
• mcpgateway/services/prompt_service.py - Prompt metrics logic
• mcpgateway/services/server_service.py - Server metrics logic
• mcpgateway/db.py - Metric model properties
• mcpgateway/schemas.py - API response schemas

Database Models

• ToolMetric - Tool execution metrics
• ResourceMetric - Resource access metrics
• PromptMetric - Prompt usage metrics
• ServerMetric - Server interaction metrics

Notes

• The UI should continue to display top 5 for clarity and performance
• Export functionality should provide complete data for analysis
• Consider adding a "View All" link for each metric category in future iterations
• Response time truncation should be consistent across all displays