dataflow.mcp.health¶

MCP Health Monitoring for haive-dataflow.

This module provides health monitoring and management capabilities for MCP servers, including connection status tracking, performance metrics, and automatic recovery.

Classes:: MCPHealthMonitor: Main health monitoring service MCPHealthChecker: Individual server health checker

Classes¶

`MCPHealthChecker`	Health checker for individual MCP servers.
`MCPHealthMonitor`	Health monitoring service for MCP servers.

Module Contents¶

class dataflow.mcp.health.MCPHealthChecker(server_name, server_config)¶

Health checker for individual MCP servers.

This class handles health checking for a single MCP server including connection testing, response time measurement, and recovery attempts.

Initialize health checker.

Parameters:

server_name (str) – Name of the server to monitor
server_config (haive.dataflow.registry.models.MCPServerConfig) – Server configuration

async attempt_recovery()¶: Attempt to recover the server connection.

async check_health()¶

Perform health check on the server.

Returns:: Current health status
Return type:: haive.dataflow.registry.models.MCPServerHealth

class dataflow.mcp.health.MCPHealthMonitor(mcp_client=None, monitoring_interval=30)¶

Health monitoring service for MCP servers.

This class provides comprehensive health monitoring for MCP servers including: - Periodic health checks - Performance metric tracking - Automatic recovery attempts - Health status reporting

mcp_client¶: Reference to the MCP client

health_checkers¶: Dictionary of server health checkers

monitoring_interval¶: Seconds between health checks

is_monitoring¶: Whether monitoring is currently active

Examples

monitor = MCPHealthMonitor(mcp_client) await monitor.start_monitoring()

# Get health status status = await monitor.get_health_summary() print(f”Healthy servers: {status[‘healthy_count’]}”)

Initialize health monitor.

Parameters:

mcp_client – MCP client instance to monitor
monitoring_interval (int) – Seconds between health checks

async check_all_servers()¶

Perform health check on all servers.

Returns:: Dictionary of server name to health status
Return type:: dict[str, haive.dataflow.registry.models.MCPServerHealth]

async get_health_summary()¶

Get summary of health status across all servers.

Returns:: Summary dictionary with health metrics
Return type:: dict[str, any]

async recover_failed_servers()¶

Attempt to recover failed servers.

Returns:: List of server names that were successfully recovered
Return type:: list[str]

async start_monitoring()¶: Start health monitoring for all connected servers.

async stop_monitoring()¶: Stop health monitoring.