SigNoz Health Check

Perform a comprehensive health check of the SigNoz observability platform, analyzing services, logs, metrics, traces, and alerts to identify highlights and issues.

Usage

/signoz-health-check [timeRange]

Examples:

• /signoz-health-check -- Check last 24 hours (default)
• /signoz-health-check 1h -- Check last 1 hour
• /signoz-health-check 6h -- Check last 6 hours
• /signoz-health-check 7d -- Check last 7 days

What It Checks

1. Service Health

• List all active services
• Call rates and request volumes
• Error rates (identify services with >0% errors)
• P99 latency metrics
• Top operations per service

2. Error Analysis

• Recent ERROR/FATAL logs across all services
• Error patterns and frequency
• Services most affected by errors
• Common error messages

3. Traces

• Error traces for high-error services
• Slow traces (P99 > threshold)
• Trace patterns and bottlenecks

4. Alerts

• Active alerts currently firing
• Alert frequency and patterns
• Services with no alerts configured (risk)

5. Metrics

• Available metric keys
• Key performance indicators
• Resource utilization patterns

6. Dashboards

• List of configured dashboards
• Coverage gaps

Output Format

The check produces a structured report with:

Highlights

• ✅ Well-performing services (0% error rate)
• 📊 High traffic endpoints
• 🎯 Key metrics and thresholds met

Issues

• 🔴 Critical: Services with high error rates (>1%)
• 🟡 Warning: Recurring errors or performance degradation
• 🟠 Attention: Missing alerts or monitoring gaps

Recommendations

• Immediate actions required
• Monitoring improvements needed
• Performance optimizations to consider

Time Range Format

Supported formats:

• 30m, 1h, 2h, 6h -- Hours/minutes
• 24h, 7d -- Days
• Default: 24h (24 hours)

Prerequisites

• SigNoz MCP server must be configured and connected
• Access to SigNoz API required

Example Output

# SigNoz Health Check - Last 24 Hours

## 📊 System Overview
- 5 active services
- 141,588 total requests
- 103 errors (0.07% overall)

## ✅ Highlights
- api: 103,878 calls, 0% errors
- ws-server: 13,268 calls, 0% errors
- 6 dashboards configured

## ⚠️ Issues
🔴 code-api: 1.98% error rate (103/5,200 calls)
   - P99 latency: 56.8s
   - Error: Session errors on POST /v1/sessions/:sessionId/v1/messages

🟡 Recurring errors:
   - "Test case not found" - 30+ occurrences
   - "Failed to ensure TTLs on job tracker keys" - Every 5 minutes

## 💡 Recommendations
1. Investigate code-api session errors immediately
2. Create alerts for error rate > 1%
3. Fix Redis TTL management issues

How It Works

1. Query services: Get all active services for the time range
2. Check alerts: List active and historical alerts
3. Analyze logs: Query error logs (ERROR/FATAL severity)
4. Inspect traces: Sample error traces from high-error services
5. Review metrics: Check available metrics and dashboards
6. Synthesize: Combine data into actionable insights

Common Issues

"No services found"

• Check SigNoz is running and instrumentation is active
• Verify the time range has data

"MCP server not connected"

• Ensure SigNoz MCP server is configured in Claude settings
• Check network connectivity to SigNoz

"Incomplete data"

• Some services may not have full instrumentation
• Check for dataWarning fields indicating overflow

Integration with Other Skills

This skill complements:

• Investigation workflows (drill into specific errors)
• Alerting setup (identify missing alerts)
• Performance optimization (find bottlenecks)

Key Metrics Tracked

Metric	Threshold	Action
Error rate	>1%	Critical investigation
P99 latency	>10s	Performance review
Alert count	0 configured	Add monitoring
Recurring errors	>10/hour	Root cause analysis

When to Use

• Regular health checks: Run daily or weekly
• Incident investigation: Start with this to get context
• Performance reviews: Identify optimization opportunities
• Capacity planning: Understand traffic patterns
• Post-deployment: Verify system health after changes

Advanced Usage

The skill collects data that can be further analyzed:

• Correlate error spikes with deployments
• Track error rate trends over time
• Identify cascade failures across services
• Monitor specific error patterns

For deeper investigation of specific issues, use:

• mcp__signoz__signoz_get_trace_details for trace analysis
• mcp__signoz__signoz_search_logs_by_service for detailed logs
• mcp__signoz__signoz_get_alert_history for alert patterns