redis-benchmarks-specification 0.1.274__py3-none-any.whl → 0.1.275__py3-none-any.whl

redis_benchmarks_specification/__common__/timeseries.py

@@ -133,6 +133,9 @@ def extract_results_table(
                 use_metric_context_path = False
                 if len(find_res) > 1:
                     use_metric_context_path = True
+                # Always use context path for precision_summary metrics to show actual precision levels
+                if "precision_summary" in metric_jsonpath and "*" in metric_jsonpath:
+                    use_metric_context_path = True
                 for metric in find_res:
                     metric_name = str(metric.path)
                     metric_value = float(metric.value)
@@ -142,15 +145,34 @@ def extract_results_table(
                     if metric_jsonpath[0] == ".":
                         metric_jsonpath = metric_jsonpath[1:]
 
+                    # For precision_summary metrics, construct the full resolved path for display
+                    display_path = metric_jsonpath
+                    if "precision_summary" in metric_jsonpath and "*" in metric_jsonpath and use_metric_context_path:
+                        # Replace the wildcard with the actual precision level
+                        display_path = metric_jsonpath.replace("*", metric_context_path)
+
                     # retro-compatible naming
                     if use_metric_context_path is False:
                         metric_name = metric_jsonpath
-
-                    metric_name = metric_name.replace("'", "")
-                    metric_name = metric_name.replace('"', "")
-                    metric_name = metric_name.replace("(", "")
-                    metric_name = metric_name.replace(")", "")
-                    metric_name = metric_name.replace(" ", "_")
+                    else:
+                        # For display purposes, use the resolved path for precision_summary
+                        if "precision_summary" in metric_jsonpath and "*" in metric_jsonpath:
+                            metric_name = display_path
+                        else:
+                            # Clean up the metric name for other cases
+                            metric_name = metric_name.replace("'", "")
+                            metric_name = metric_name.replace('"', "")
+                            metric_name = metric_name.replace("(", "")
+                            metric_name = metric_name.replace(")", "")
+                            metric_name = metric_name.replace(" ", "_")
+
+                    # Apply standard cleaning to all metric names
+                    if not ("precision_summary" in metric_jsonpath and "*" in metric_jsonpath and use_metric_context_path):
+                        metric_name = metric_name.replace("'", "")
+                        metric_name = metric_name.replace('"', "")
+                        metric_name = metric_name.replace("(", "")
+                        metric_name = metric_name.replace(")", "")
+                        metric_name = metric_name.replace(" ", "_")
 
                     results_matrix.append(
                         [

redis_benchmarks_specification/__runner__/args.py

@@ -225,4 +225,40 @@ def create_client_runner_args(project_name):
         default="",
         help="UNIX Domain socket name",
     )
+    parser.add_argument(
+        "--enable-remote-profiling",
+        default=False,
+        action="store_true",
+        help="Enable remote profiling of Redis processes via HTTP GET endpoint. Profiles are collected in folded format during benchmark execution.",
+    )
+    parser.add_argument(
+        "--remote-profile-host",
+        type=str,
+        default="localhost",
+        help="Host for remote profiling HTTP endpoint. Default is localhost.",
+    )
+    parser.add_argument(
+        "--remote-profile-port",
+        type=int,
+        default=8080,
+        help="Port for remote profiling HTTP endpoint. Default is 8080.",
+    )
+    parser.add_argument(
+        "--remote-profile-output-dir",
+        type=str,
+        default="profiles",
+        help="Directory to store remote profiling output files. Default is 'profiles/'.",
+    )
+    parser.add_argument(
+        "--remote-profile-username",
+        type=str,
+        default=None,
+        help="Username for HTTP basic authentication to remote profiling endpoint. Optional.",
+    )
+    parser.add_argument(
+        "--remote-profile-password",
+        type=str,
+        default=None,
+        help="Password for HTTP basic authentication to remote profiling endpoint. Optional.",
+    )
     return parser

redis_benchmarks_specification/__runner__/remote_profiling.py

@@ -0,0 +1,329 @@
+"""
+Remote profiling utilities for Redis benchmark runner.
+
+This module provides functionality to trigger remote profiling of Redis processes
+via HTTP GET endpoints during benchmark execution. Profiles are collected in
+folded format for performance analysis.
+"""
+
+import datetime
+import logging
+import os
+import threading
+import time
+from pathlib import Path
+from typing import Optional, Dict, Any
+import requests
+
+
+def extract_redis_pid(redis_conn) -> Optional[int]:
+    """
+    Extract Redis process ID from Redis INFO command.
+    
+    Args:
+        redis_conn: Redis connection object
+        
+    Returns:
+        Redis process ID as integer, or None if not found
+    """
+    try:
+        redis_info = redis_conn.info()
+        pid = redis_info.get("process_id")
+        if pid is not None:
+            logging.info(f"Extracted Redis PID: {pid}")
+            return int(pid)
+        else:
+            logging.warning("Redis process_id not found in INFO command")
+            return None
+    except Exception as e:
+        logging.error(f"Failed to extract Redis PID: {e}")
+        return None
+
+
+def extract_redis_metadata(redis_conn) -> Dict[str, Any]:
+    """
+    Extract Redis metadata for profile comments.
+    
+    Args:
+        redis_conn: Redis connection object
+        
+    Returns:
+        Dictionary containing Redis metadata
+    """
+    try:
+        redis_info = redis_conn.info()
+        metadata = {
+            "redis_version": redis_info.get("redis_version", "unknown"),
+            "redis_git_sha1": redis_info.get("redis_git_sha1", "unknown"),
+            "redis_git_dirty": redis_info.get("redis_git_dirty", "unknown"),
+            "redis_build_id": redis_info.get("redis_build_id", "unknown"),
+            "process_id": redis_info.get("process_id", "unknown"),
+            "tcp_port": redis_info.get("tcp_port", "unknown"),
+        }
+        
+        # Use build_id if git_sha1 is empty or 0
+        if metadata["redis_git_sha1"] in ("", 0, "0"):
+            metadata["redis_git_sha1"] = metadata["redis_build_id"]
+            
+        logging.info(f"Extracted Redis metadata: version={metadata['redis_version']}, sha={metadata['redis_git_sha1']}, pid={metadata['process_id']}")
+        return metadata
+    except Exception as e:
+        logging.error(f"Failed to extract Redis metadata: {e}")
+        return {
+            "redis_version": "unknown",
+            "redis_git_sha1": "unknown", 
+            "redis_git_dirty": "unknown",
+            "redis_build_id": "unknown",
+            "process_id": "unknown",
+            "tcp_port": "unknown",
+        }
+
+
+def calculate_profile_duration(benchmark_duration_seconds: int) -> int:
+    """
+    Calculate profiling duration based on benchmark duration.
+    
+    Args:
+        benchmark_duration_seconds: Expected benchmark duration in seconds
+        
+    Returns:
+        Profiling duration in seconds (minimum: benchmark duration, maximum: 30)
+    """
+    # Minimum duration is the benchmark duration, maximum is 30 seconds
+    duration = min(max(benchmark_duration_seconds, 10), 30)
+    logging.info(f"Calculated profile duration: {duration}s (benchmark: {benchmark_duration_seconds}s)")
+    return duration
+
+
+def trigger_remote_profile(
+    host: str,
+    port: int,
+    pid: int,
+    duration: int,
+    timeout: int = 60,
+    username: Optional[str] = None,
+    password: Optional[str] = None
+) -> Optional[str]:
+    """
+    Trigger remote profiling via HTTP GET request.
+
+    Args:
+        host: Remote host address
+        port: Remote port number
+        pid: Redis process ID
+        duration: Profiling duration in seconds
+        timeout: HTTP request timeout in seconds
+        username: Optional username for HTTP basic authentication
+        password: Optional password for HTTP basic authentication
+
+    Returns:
+        Profile content in folded format, or None if failed
+    """
+    url = f"http://{host}:{port}/debug/folded/profile"
+    params = {
+        "pid": pid,
+        "seconds": duration
+    }
+
+    # Prepare authentication if provided
+    auth = None
+    if username is not None and password is not None:
+        auth = (username, password)
+        logging.info(f"Using HTTP basic authentication with username: {username}")
+
+    try:
+        logging.info(f"Triggering remote profile: {url} with PID={pid}, duration={duration}s")
+        response = requests.get(url, params=params, timeout=timeout, auth=auth)
+        response.raise_for_status()
+
+        profile_content = response.text
+        logging.info(f"Successfully collected profile: {len(profile_content)} characters")
+        return profile_content
+
+    except requests.exceptions.Timeout:
+        logging.error(f"Remote profiling request timed out after {timeout}s")
+        return None
+    except requests.exceptions.ConnectionError:
+        logging.error(f"Failed to connect to remote profiling endpoint: {host}:{port}")
+        return None
+    except requests.exceptions.HTTPError as e:
+        logging.error(f"HTTP error during remote profiling: {e}")
+        if e.response.status_code == 401:
+            logging.error("Authentication failed - check username and password")
+        return None
+    except Exception as e:
+        logging.error(f"Unexpected error during remote profiling: {e}")
+        return None
+
+
+def save_profile_with_metadata(
+    profile_content: str,
+    benchmark_name: str,
+    output_dir: str,
+    redis_metadata: Dict[str, Any],
+    duration: int
+) -> Optional[str]:
+    """
+    Save profile content to file with metadata comments.
+    
+    Args:
+        profile_content: Profile data in folded format
+        benchmark_name: Name of the benchmark
+        output_dir: Output directory path
+        redis_metadata: Redis metadata dictionary
+        duration: Profiling duration in seconds
+        
+    Returns:
+        Path to saved file, or None if failed
+    """
+    try:
+        # Create output directory if it doesn't exist
+        Path(output_dir).mkdir(parents=True, exist_ok=True)
+        
+        # Generate filename
+        filename = f"{benchmark_name}.folded"
+        filepath = os.path.join(output_dir, filename)
+        
+        # Generate timestamp
+        timestamp = datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S")
+        
+        # Create metadata comment
+        metadata_comment = (
+            f"# profile from redis sha = {redis_metadata['redis_git_sha1']} "
+            f"and pid {redis_metadata['process_id']} for duration of {duration}s. "
+            f"collection in date {timestamp}\n"
+            f"# redis_version: {redis_metadata['redis_version']}\n"
+            f"# redis_git_dirty: {redis_metadata['redis_git_dirty']}\n"
+            f"# tcp_port: {redis_metadata['tcp_port']}\n"
+        )
+        
+        # Write file with metadata and profile content
+        with open(filepath, 'w') as f:
+            f.write(metadata_comment)
+            f.write(profile_content)
+            
+        logging.info(f"Saved profile to: {filepath}")
+        return filepath
+        
+    except Exception as e:
+        logging.error(f"Failed to save profile file: {e}")
+        return None
+
+
+class RemoteProfiler:
+    """
+    Remote profiler class to handle threaded profiling execution.
+    """
+
+    def __init__(self, host: str, port: int, output_dir: str, username: Optional[str] = None, password: Optional[str] = None):
+        self.host = host
+        self.port = port
+        self.output_dir = output_dir
+        self.username = username
+        self.password = password
+        self.profile_thread = None
+        self.profile_result = None
+        self.profile_error = None
+        
+    def start_profiling(
+        self,
+        redis_conn,
+        benchmark_name: str,
+        benchmark_duration_seconds: int
+    ) -> bool:
+        """
+        Start profiling in a separate thread.
+        
+        Args:
+            redis_conn: Redis connection object
+            benchmark_name: Name of the benchmark
+            benchmark_duration_seconds: Expected benchmark duration
+            
+        Returns:
+            True if profiling thread started successfully, False otherwise
+        """
+        try:
+            # Extract Redis metadata and PID
+            redis_metadata = extract_redis_metadata(redis_conn)
+            pid = redis_metadata.get("process_id")
+            
+            if pid == "unknown" or pid is None:
+                logging.error("Cannot start remote profiling: Redis PID not available")
+                return False
+                
+            # Calculate profiling duration
+            duration = calculate_profile_duration(benchmark_duration_seconds)
+            
+            # Start profiling thread
+            self.profile_thread = threading.Thread(
+                target=self._profile_worker,
+                args=(pid, duration, benchmark_name, redis_metadata),
+                daemon=True
+            )
+            self.profile_thread.start()
+            
+            logging.info(f"Started remote profiling thread for benchmark: {benchmark_name}")
+            return True
+            
+        except Exception as e:
+            logging.error(f"Failed to start remote profiling: {e}")
+            return False
+    
+    def _profile_worker(self, pid: int, duration: int, benchmark_name: str, redis_metadata: Dict[str, Any]):
+        """
+        Worker function for profiling thread.
+        """
+        try:
+            # Trigger remote profiling
+            profile_content = trigger_remote_profile(
+                self.host, self.port, pid, duration,
+                username=self.username, password=self.password
+            )
+
+            if profile_content:
+                # Save profile with metadata
+                filepath = save_profile_with_metadata(
+                    profile_content, benchmark_name, self.output_dir, redis_metadata, duration
+                )
+                self.profile_result = filepath
+            else:
+                self.profile_error = "Failed to collect profile content"
+
+        except Exception as e:
+            self.profile_error = f"Profile worker error: {e}"
+            logging.error(self.profile_error)
+    
+    def wait_for_completion(self, timeout: int = 60) -> bool:
+        """
+        Wait for profiling thread to complete.
+        
+        Args:
+            timeout: Maximum time to wait in seconds
+            
+        Returns:
+            True if completed successfully, False if timed out or failed
+        """
+        if self.profile_thread is None:
+            return False
+            
+        try:
+            self.profile_thread.join(timeout=timeout)
+            
+            if self.profile_thread.is_alive():
+                logging.warning(f"Remote profiling thread did not complete within {timeout}s")
+                return False
+                
+            if self.profile_error:
+                logging.error(f"Remote profiling failed: {self.profile_error}")
+                return False
+                
+            if self.profile_result:
+                logging.info(f"Remote profiling completed successfully: {self.profile_result}")
+                return True
+            else:
+                logging.warning("Remote profiling completed but no result available")
+                return False
+                
+        except Exception as e:
+            logging.error(f"Error waiting for remote profiling completion: {e}")
+            return False