Reduces MCP response from ~12.3k tokens (~1000 lines) to only active servers. Empty pool slots (MAINT with no health check) are filtered by default. Added show_all parameter to optionally show all slots. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
225 lines
8.9 KiB
Python
225 lines
8.9 KiB
Python
"""Health check tools for HAProxy MCP Server."""
|
|
|
|
import json
|
|
import subprocess
|
|
import time
|
|
from typing import Annotated, Any
|
|
|
|
from pydantic import Field
|
|
|
|
from ..config import (
|
|
MAP_FILE,
|
|
SERVERS_FILE,
|
|
DB_FILE,
|
|
HAPROXY_CONTAINER,
|
|
)
|
|
from ..exceptions import HaproxyError
|
|
from ..validation import validate_domain, validate_backend_name
|
|
from ..haproxy_client import haproxy_cmd
|
|
from ..file_ops import get_backend_and_prefix
|
|
from ..utils import parse_stat_csv, parse_servers_state
|
|
from ..ssh_ops import run_command, remote_file_exists
|
|
from ..config import REMOTE_MODE
|
|
|
|
|
|
def register_health_tools(mcp):
|
|
"""Register health check tools with MCP server."""
|
|
|
|
@mcp.tool()
|
|
def haproxy_health() -> str:
|
|
"""Check overall system health (MCP server, HAProxy, config files).
|
|
|
|
Use this for monitoring integration. Returns "healthy" if all components are OK.
|
|
|
|
Returns:
|
|
JSON with:
|
|
- status: "healthy" or "unhealthy"
|
|
- components.mcp: MCP server status
|
|
- components.haproxy: HAProxy connectivity, version, uptime
|
|
- components.config_files: map_file and servers_file accessibility
|
|
|
|
Example:
|
|
# Returns: {"status": "healthy", "components": {"mcp": {"status": "ok"}, ...}}
|
|
"""
|
|
result: dict[str, Any] = {
|
|
"status": "healthy",
|
|
"timestamp": time.time(),
|
|
"components": {
|
|
"mcp": {"status": "ok"},
|
|
"haproxy": {"status": "unknown"},
|
|
"config_files": {"status": "unknown"}
|
|
}
|
|
}
|
|
|
|
# Check HAProxy connectivity
|
|
try:
|
|
info = haproxy_cmd("show info")
|
|
for line in info.split("\n"):
|
|
if line.startswith("Version:"):
|
|
result["components"]["haproxy"]["version"] = line.split(":", 1)[1].strip()
|
|
elif line.startswith("Uptime_sec:"):
|
|
result["components"]["haproxy"]["uptime_sec"] = int(line.split(":", 1)[1].strip())
|
|
result["components"]["haproxy"]["status"] = "ok"
|
|
except HaproxyError as e:
|
|
result["components"]["haproxy"]["status"] = "error"
|
|
result["components"]["haproxy"]["error"] = str(e)
|
|
result["status"] = "degraded"
|
|
|
|
# Check container status
|
|
try:
|
|
container_result = run_command(
|
|
["podman", "inspect", "--format", "{{.State.Status}}", HAPROXY_CONTAINER],
|
|
timeout=5,
|
|
)
|
|
if container_result.returncode == 0:
|
|
container_status = container_result.stdout.strip()
|
|
result["components"]["container"] = {
|
|
"status": "ok" if container_status == "running" else container_status,
|
|
"state": container_status
|
|
}
|
|
else:
|
|
result["components"]["container"] = {"status": "error", "error": container_result.stderr.strip()}
|
|
result["status"] = "unhealthy"
|
|
except subprocess.TimeoutExpired:
|
|
result["components"]["container"] = {"status": "timeout"}
|
|
result["status"] = "unhealthy"
|
|
except (OSError, subprocess.SubprocessError) as e:
|
|
result["components"]["container"] = {"status": "error", "error": str(e)}
|
|
|
|
# Check configuration files
|
|
files_ok = True
|
|
file_status: dict[str, str] = {}
|
|
for name, path in [("map_file", MAP_FILE), ("db_file", DB_FILE)]:
|
|
# DB file is always local (SQLite runs in MCP container), map_file may be remote
|
|
if name == "db_file" or not REMOTE_MODE:
|
|
exists = __import__('os').path.exists(path)
|
|
else:
|
|
exists = remote_file_exists(path)
|
|
if exists:
|
|
file_status[name] = "ok"
|
|
else:
|
|
file_status[name] = "missing"
|
|
files_ok = False
|
|
|
|
result["components"]["config_files"]["files"] = file_status
|
|
result["components"]["config_files"]["status"] = "ok" if files_ok else "warning"
|
|
if not files_ok:
|
|
result["status"] = "degraded"
|
|
|
|
return json.dumps(result, indent=2)
|
|
|
|
@mcp.tool()
|
|
def haproxy_domain_health(
|
|
domain: Annotated[str, Field(description="Domain name to check health for (e.g., api.example.com)")]
|
|
) -> str:
|
|
"""Check health status of backend servers for a specific domain.
|
|
|
|
Returns JSON with status (healthy/degraded/down/no_servers), server list, and counts.
|
|
"""
|
|
if not validate_domain(domain):
|
|
return json.dumps({"error": "Invalid domain format"})
|
|
|
|
try:
|
|
backend, _ = get_backend_and_prefix(domain)
|
|
except ValueError as e:
|
|
return json.dumps({"error": str(e)})
|
|
|
|
result: dict[str, Any] = {
|
|
"domain": domain,
|
|
"backend": backend,
|
|
"status": "unknown",
|
|
"servers": [],
|
|
"healthy_count": 0,
|
|
"total_count": 0
|
|
}
|
|
|
|
try:
|
|
# Get server states
|
|
state_output = haproxy_cmd("show servers state")
|
|
stat_output = haproxy_cmd("show stat")
|
|
|
|
# Build status map from stat output (has UP/DOWN/MAINT status)
|
|
status_map: dict[str, dict[str, str]] = {}
|
|
for stat in parse_stat_csv(stat_output):
|
|
if stat["pxname"] == backend and stat["svname"] not in ["FRONTEND", "BACKEND"]:
|
|
status_map[stat["svname"]] = {
|
|
"status": stat["status"],
|
|
"check_status": stat["check_status"],
|
|
"weight": stat["weight"]
|
|
}
|
|
|
|
# Parse server state for address info
|
|
parsed_state = parse_servers_state(state_output)
|
|
backend_servers = parsed_state.get(backend, {})
|
|
|
|
for server_name, srv_info in backend_servers.items():
|
|
addr = srv_info["addr"]
|
|
port = srv_info["port"]
|
|
|
|
# Skip disabled servers (0.0.0.0)
|
|
if addr == "0.0.0.0":
|
|
continue
|
|
|
|
server_info: dict[str, Any] = {
|
|
"name": server_name,
|
|
"addr": f"{addr}:{port}",
|
|
"status": "unknown"
|
|
}
|
|
|
|
# Get status from stat output
|
|
if server_name in status_map:
|
|
server_info["status"] = status_map[server_name]["status"]
|
|
server_info["check_status"] = status_map[server_name]["check_status"]
|
|
server_info["weight"] = status_map[server_name]["weight"]
|
|
|
|
result["servers"].append(server_info)
|
|
result["total_count"] += 1
|
|
|
|
if server_info["status"] == "UP":
|
|
result["healthy_count"] += 1
|
|
|
|
# Determine overall status
|
|
if result["total_count"] == 0:
|
|
result["status"] = "no_servers"
|
|
elif result["healthy_count"] == result["total_count"]:
|
|
result["status"] = "healthy"
|
|
elif result["healthy_count"] > 0:
|
|
result["status"] = "degraded"
|
|
else:
|
|
result["status"] = "down"
|
|
|
|
return json.dumps(result, indent=2)
|
|
|
|
except HaproxyError as e:
|
|
return json.dumps({"error": str(e)})
|
|
|
|
@mcp.tool()
|
|
def haproxy_get_server_health(
|
|
backend: Annotated[str, Field(default="", description="Optional: Backend name to filter (e.g., 'pool_1'). Empty = all backends")] = "",
|
|
show_all: Annotated[bool, Field(default=False, description="Show all slots including empty MAINT slots. Default: only active/configured servers")] = False,
|
|
) -> str:
|
|
"""Get health status of all servers (low-level view). For domain-specific, use haproxy_domain_health.
|
|
|
|
By default, hides empty pool slots (MAINT with no health check) to reduce output size.
|
|
Use show_all=True to see all slots including unconfigured ones.
|
|
"""
|
|
if backend and not validate_backend_name(backend):
|
|
return "Error: Invalid backend name (use alphanumeric, underscore, hyphen only)"
|
|
try:
|
|
result = haproxy_cmd("show stat")
|
|
servers = []
|
|
for stat in parse_stat_csv(result):
|
|
if stat["svname"] not in ["FRONTEND", "BACKEND", ""]:
|
|
if backend and stat["pxname"] != backend:
|
|
continue
|
|
# Skip empty pool slots (MAINT with no health check = unconfigured)
|
|
if not show_all and stat["status"] == "MAINT" and not stat["check_status"]:
|
|
continue
|
|
servers.append(
|
|
f"• {stat['pxname']}/{stat['svname']}: {stat['status']} "
|
|
f"(weight: {stat['weight']}, check: {stat['check_status']})"
|
|
)
|
|
return "\n".join(servers) if servers else "No servers found"
|
|
except HaproxyError as e:
|
|
return f"Error: {e}"
|