mcp-hydrolix

import logging import json from typing import Optional, List, Any import concurrent.futures import atexit import clickhouse_connect from clickhouse_connect.driver.binding import format_query_value from dotenv import load_dotenv from fastmcp import FastMCP from fastmcp.exceptions import ToolError from dataclasses import dataclass, field, asdict, is_dataclass from starlette.requests import Request from starlette.responses import PlainTextResponse from mcp_hydrolix.mcp_env import get_config @dataclass class Column: database: str table: str name: str column_type: str default_kind: Optional[str] default_expression: Optional[str] comment: Optional[str] @dataclass class Table: database: str name: str engine: str create_table_query: str dependencies_database: str dependencies_table: str engine_full: str sorting_key: str primary_key: str total_rows: int total_bytes: int total_bytes_uncompressed: int parts: int active_parts: int total_marks: int comment: Optional[str] = None columns: List[Column] = field(default_factory=list) MCP_SERVER_NAME = "mcp-hydrolix" # Configure logging logging.basicConfig( level=logging.INFO, format="%(asctime)s - %(name)s - %(levelname)s - %(message)s" ) logger = logging.getLogger(MCP_SERVER_NAME) QUERY_EXECUTOR = concurrent.futures.ThreadPoolExecutor(max_workers=10) atexit.register(lambda: QUERY_EXECUTOR.shutdown(wait=True)) SELECT_QUERY_TIMEOUT_SECS = 30 load_dotenv() mcp = FastMCP( name=MCP_SERVER_NAME, dependencies=[ "clickhouse-connect", "python-dotenv", "pip-system-certs", ], ) @mcp.custom_route("/health", methods=["GET"]) async def health_check(request: Request) -> PlainTextResponse: """Health check endpoint for monitoring server status. Returns OK if the server is running and can connect to Hydrolix. """ try: # Try to create a client connection to verify query-head connectivity client = create_hydrolix_client() version = client.server_version return PlainTextResponse(f"OK - Connected to Hydrolix compatible with ClickHouse {version}") except Exception as e: # Return 503 Service Unavailable if we can't connect to Hydrolix return PlainTextResponse(f"ERROR - Cannot connect to Hydrolix: {str(e)}", status_code=503) def result_to_table(query_columns, result) -> List[Table]: return [Table(**dict(zip(query_columns, row))) for row in result] def result_to_column(query_columns, result) -> List[Column]: return [Column(**dict(zip(query_columns, row))) for row in result] def to_json(obj: Any) -> str: if is_dataclass(obj): return json.dumps(asdict(obj), default=to_json) elif isinstance(obj, list): return [to_json(item) for item in obj] elif isinstance(obj, dict): return {key: to_json(value) for key, value in obj.items()} return obj @mcp.tool() def list_databases(): """List available Hydrolix databases""" logger.info("Listing all databases") client = create_hydrolix_client() result = client.command("SHOW DATABASES") # Convert newline-separated string to list and trim whitespace if isinstance(result, str): databases = [db.strip() for db in result.strip().split("\n")] else: databases = [result] logger.info(f"Found {len(databases)} databases") return json.dumps(databases) @mcp.tool() def list_tables(database: str, like: Optional[str] = None, not_like: Optional[str] = None): """List available Hydrolix tables in a database, including schema, comment, row count, and column count.""" logger.info(f"Listing tables in database '{database}'") client = create_hydrolix_client() query = f"SELECT database, name, engine, create_table_query, dependencies_database, dependencies_table, engine_full, sorting_key, primary_key, total_rows, total_bytes, total_bytes_uncompressed, parts, active_parts, total_marks, comment FROM system.tables WHERE database = {format_query_value(database)}" if like: query += f" AND name LIKE {format_query_value(like)}" if not_like: query += f" AND name NOT LIKE {format_query_value(not_like)}" result = client.query(query) # Deserialize result as Table dataclass instances tables = result_to_table(result.column_names, result.result_rows) for table in tables: column_data_query = f"SELECT database, table, name, type AS column_type, default_kind, default_expression, comment FROM system.columns WHERE database = {format_query_value(database)} AND table = {format_query_value(table.name)}" column_data_query_result = client.query(column_data_query) table.columns = [ c for c in result_to_column( column_data_query_result.column_names, column_data_query_result.result_rows, ) ] logger.info(f"Found {len(tables)} tables") return [asdict(table) for table in tables] def execute_query(query: str): client = create_hydrolix_client() try: res = client.query( query, settings={ "readonly": 1, "hdx_query_max_execution_time": SELECT_QUERY_TIMEOUT_SECS, "hdx_query_max_attempts": 1, "hdx_query_max_result_rows": 100_000, "hdx_query_max_memory_usage": 2 * 1024 * 1024 * 1024, # 2GiB "hdx_query_admin_comment": f"User: {MCP_SERVER_NAME}", }, ) logger.info(f"Query returned {len(res.result_rows)} rows") return {"columns": res.column_names, "rows": res.result_rows} except Exception as err: logger.error(f"Error executing query: {err}") raise ToolError(f"Query execution failed: {str(err)}") @mcp.tool() def run_select_query(query: str): """Run a SELECT query in a Hydrolix time-series database using the Clickhouse SQL dialect. Queries run using this tool will timeout after 30 seconds. The primary key on tables queried this way is always a timestamp. Queries should include either a LIMIT clause or a filter based on the primary key as a performance guard to ensure they return in a reasonable amount of time. Queries should select specific fields and avoid the use of SELECT * to avoid performance issues. The performance guard used for the query should be clearly communicated with the user, and the user should be informed that the query may take a long time to run if the performance guard is not used. When choosing a performance guard, the user's preference should be requested and used if available. When using aggregations, the performance guard should take form of a primary key filter, or else the LIMIT should be applied in a subquery before applying the aggregations. When matching columns based on substrings, prefix or suffix matches should be used instead of full-text search whenever possible. When searching for substrings, the syntax `column LIKE '%suffix'` or `column LIKE 'prefix%'` should be used. Example query. Purpose: get logs from the `application.logs` table. Primary key: `timestamp`. Performance guard: 10 minute recency filter. `SELECT message, timestamp FROM application.logs WHERE timestamp > now() - INTERVAL 10 MINUTES` Example query. Purpose: get the median humidity from the `weather.measurements` table. Primary key: `date`. Performance guard: 1000 row limit, applied before aggregation. `SELECT median(humidity) FROM (SELECT humidity FROM weather.measurements LIMIT 1000)` Example query. Purpose: get the lowest temperature from the `weather.measurements` table over the last 10 years. Primary key: `date`. Performance guard: date range filter. `SELECT min(temperature) FROM weather.measurements WHERE date > now() - INTERVAL 10 YEARS` Example query. Purpose: get the app name with the most log messages from the `application.logs` table in the window between new year and valentine's day of 2024. Primary key: `timestamp`. Performance guard: date range filter. `SELECT app, count(*) FROM application.logs WHERE timestamp > '2024-01-01' AND timestamp < '2024-02-14' GROUP BY app ORDER BY count(*) DESC LIMIT 1` """ logger.info(f"Executing SELECT query: {query}") try: future = QUERY_EXECUTOR.submit(execute_query, query) try: result = future.result(timeout=SELECT_QUERY_TIMEOUT_SECS) # Check if we received an error structure from execute_query if isinstance(result, dict) and "error" in result: logger.warning(f"Query failed: {result['error']}") # MCP requires structured responses; string error messages can cause # serialization issues leading to BrokenResourceError return { "status": "error", "message": f"Query failed: {result['error']}", } return result except concurrent.futures.TimeoutError: logger.warning(f"Query timed out after {SELECT_QUERY_TIMEOUT_SECS} seconds: {query}") future.cancel() raise ToolError(f"Query timed out after {SELECT_QUERY_TIMEOUT_SECS} seconds") except ToolError: raise except Exception as e: logger.error(f"Unexpected error in run_select_query: {str(e)}") raise RuntimeError(f"Unexpected error during query execution: {str(e)}") def create_hydrolix_client(): client_config = get_config().get_client_config() auth_info = ( f"as {client_config['username']}" if "username" in client_config else "using service account token" ) logger.info( f"Creating Hydrolix client connection to {client_config['host']}:{client_config['port']} " f"{auth_info} " f"(secure={client_config['secure']}, verify={client_config['verify']}, " f"connect_timeout={client_config['connect_timeout']}s, " f"send_receive_timeout={client_config['send_receive_timeout']}s)" ) try: client = clickhouse_connect.get_client(**client_config) # Test the connection version = client.server_version logger.info(f"Successfully connected to Hydrolix compatible with ClickHouse {version}") return client except Exception as e: logger.error(f"Failed to connect to Hydrolix: {str(e)}") raise