Error Tracking

Error tracking helps you identify, analyze, and resolve issues with your API requests. LangMart provides detailed error analytics to help you maintain high reliability.

Viewing Errors

Dashboard

Navigate to Analytics > Errors to view the error tracking dashboard:

• Error summary statistics
• Error breakdown by type
• Recent error list with details
• Suggested fixes for common issues

API Access

curl -X GET "https://api.langmart.ai/api/account/errors" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -G \
  -d "start_date=2025-01-01T00:00:00Z" \
  -d "end_date=2025-01-31T23:59:59Z"

Error Summary

Summary Statistics

{
  "summary": {
    "total_errors": 231,
    "error_rate": 1.5,
    "trend": -12.5,
    "days_with_errors": 28,
    "server_errors": 45,
    "client_errors": 186,
    "period": {
      "start": "2025-01-01T00:00:00Z",
      "end": "2025-01-31T23:59:59Z"
    }
  }
}

Understanding Summary Fields

Error Types

Error Breakdown

{
  "breakdown": [
    {
      "status_code": 429,
      "error_code": "PROVIDER_OPENAI_429",
      "error_type": "rate_limit_error",
      "count": 85,
      "percentage": 36.8,
      "last_occurrence": "2025-01-31T14:30:00Z"
    },
    {
      "status_code": 401,
      "error_code": "GATEWAY1_AUTH_001",
      "error_type": "authentication_error",
      "count": 52,
      "percentage": 22.5,
      "last_occurrence": "2025-01-31T12:15:00Z"
    },
    {
      "status_code": 400,
      "error_code": "API_VALIDATION_001",
      "error_type": "invalid_request_error",
      "count": 38,
      "percentage": 16.5,
      "last_occurrence": "2025-01-31T10:45:00Z"
    }
  ]
}

Common Error Types

Debugging Failed Requests

Error Details

Each error includes detailed information:

{
  "errors": [
    {
      "id": "req_abc123",
      "request_id": "req_abc123",
      "created_at": "2025-01-31T14:30:00Z",
      "method": "POST",
      "endpoint": "/v1/chat/completions",
      "response_status": 429,
      "error_code": "PROVIDER_OPENAI_429",
      "error_type": "rate_limit_error",
      "error_message": "Rate limit exceeded",
      "model_name": "gpt-4o",
      "provider_name": "OpenAI",
      "latency_ms": 450,
      "total_cost": null,
      "suggested_fix": "Rate limit exceeded. Try: 1) Reduce request rate, 2) Upgrade to higher tier, 3) Use a different model, or 4) Implement exponential backoff.",
      "severity": "high"
    }
  ]
}

Error Severity Levels

Filtering Errors

# Filter by error type
curl "https://api.langmart.ai/api/account/errors?error_type=rate_limit_error"

# Filter by status code
curl "https://api.langmart.ai/api/account/errors?status_code=429"

# Filter by model
curl "https://api.langmart.ai/api/account/errors?model=gpt-4o"

# Pagination
curl "https://api.langmart.ai/api/account/errors?limit=20&offset=0"

Common Error Codes

Authentication Errors (401)

Rate Limit Errors (429)

Validation Errors (400)

Not Found Errors (404)

Service Errors (500+)

Suggested Fixes

LangMart automatically provides suggested fixes based on the error:

Rate Limiting (429)

Rate limit exceeded. Try:
1) Reduce request rate
2) Upgrade to higher tier
3) Use a different model
4) Implement exponential backoff

Authentication (401)

Invalid API key. Go to Connections settings and update your API key for this provider.

Model Not Found (404)

Model not found. This model may have been removed or renamed. Try browsing available models in the Models page.

Server Errors (500+)

Server error. This is usually temporary. Try:
1) Retry the request
2) Wait a few minutes
3) Contact support if it persists

Token Limit (400)

Token limit exceeded. Try:
1) Reduce prompt length
2) Use a model with larger context window
3) Split into multiple requests

Error Trends

Monitoring Error Rate Over Time

Track how your error rate changes:

curl -X GET "https://api.langmart.ai/api/account/errors" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -G \
  -d "start_date=2025-01-01" \
  -d "end_date=2025-01-31"

The trend field shows percentage change from the previous period:

• Negative values indicate improvement
• Positive values indicate more errors

Setting Up Error Alerts

Create alerts to be notified of error spikes:

1. Navigate to Alerts
2. Create a new alert with type Error Rate
3. Set threshold (e.g., 5% error rate)
4. Choose notification channels

Best Practices

Implementing Retry Logic

For transient errors (429, 503, 504), implement exponential backoff:

import time
import random

def make_request_with_retry(request_func, max_retries=5):
    for attempt in range(max_retries):
        try:
            return request_func()
        except RateLimitError:
            if attempt == max_retries - 1:
                raise
            wait_time = (2 ** attempt) + random.uniform(0, 1)
            time.sleep(wait_time)

Reducing Rate Limit Errors

1. Request Queuing: Queue requests to stay under limits
2. Caching: Cache responses for identical requests
3. Model Diversification: Spread load across providers
4. Tier Upgrade: Consider higher API tiers

Monitoring Best Practices

1. Set Up Alerts: Get notified before issues escalate
2. Regular Review: Check error trends weekly
3. Root Cause Analysis: Investigate recurring errors
4. Documentation: Track known issues and solutions

Related Documentation

• API Errors Reference - Complete error code reference
• Connection Issues - Troubleshooting connections
• Request Logs - Detailed request analysis