kailash 0.6.6__py3-none-any.whl → 0.7.0__py3-none-any.whl

kailash/nodes/monitoring/transaction_metrics.py

@@ -0,0 +1,790 @@
+"""Transaction metrics collection and analysis node.
+
+This module provides comprehensive transaction performance monitoring with
+support for timing, aggregation, and export to various monitoring backends.
+"""
+
+import json
+import logging
+import time
+from collections import defaultdict
+from dataclasses import dataclass, field
+from datetime import UTC, datetime
+from enum import Enum
+from typing import Any, Dict, List, Optional, Tuple, Union
+
+from kailash.nodes.base import NodeParameter, register_node
+from kailash.nodes.base_async import AsyncNode
+from kailash.sdk_exceptions import NodeExecutionError
+
+logger = logging.getLogger(__name__)
+
+
+class MetricExportFormat(Enum):
+    """Supported metric export formats."""
+
+    JSON = "json"
+    PROMETHEUS = "prometheus"
+    CLOUDWATCH = "cloudwatch"
+    DATADOG = "datadog"
+    OPENTELEMETRY = "opentelemetry"
+
+
+class AggregationType(Enum):
+    """Types of metric aggregation."""
+
+    COUNT = "count"
+    SUM = "sum"
+    AVG = "avg"
+    MIN = "min"
+    MAX = "max"
+    P50 = "p50"
+    P75 = "p75"
+    P90 = "p90"
+    P95 = "p95"
+    P99 = "p99"
+    P999 = "p999"
+
+
+@dataclass
+class TransactionMetric:
+    """Represents a single transaction metric."""
+
+    transaction_id: str
+    name: str
+    start_time: float
+    end_time: Optional[float] = None
+    duration: Optional[float] = None
+    status: str = "pending"
+    error: Optional[str] = None
+    tags: Dict[str, str] = field(default_factory=dict)
+    custom_metrics: Dict[str, float] = field(default_factory=dict)
+
+
+@dataclass
+class AggregatedMetrics:
+    """Aggregated transaction metrics."""
+
+    name: str
+    count: int
+    sum_duration: float
+    min_duration: float
+    max_duration: float
+    avg_duration: float
+    percentiles: Dict[str, float]
+    success_count: int
+    error_count: int
+    error_rate: float
+    tags: Dict[str, str] = field(default_factory=dict)
+
+
+@register_node()
+class TransactionMetricsNode(AsyncNode):
+    """Node for collecting and analyzing transaction performance metrics.
+
+    This node provides comprehensive transaction monitoring including:
+    - Transaction timing and duration tracking
+    - Success/failure rate monitoring
+    - Latency percentile calculations (p50, p95, p99)
+    - Custom metric collection
+    - Multi-format export (Prometheus, CloudWatch, DataDog)
+    - Real-time and batch aggregation
+
+    Design Purpose:
+    - Enable production-grade performance monitoring
+    - Support SLA tracking and alerting
+    - Facilitate performance troubleshooting
+    - Integrate with enterprise monitoring systems
+
+    Examples:
+        >>> # Track individual transaction
+        >>> metrics_node = TransactionMetricsNode()
+        >>> result = await metrics_node.execute(
+        ...     operation="start_transaction",
+        ...     transaction_id="txn_12345",
+        ...     name="order_processing",
+        ...     tags={"region": "us-west", "customer_tier": "premium"}
+        ... )
+
+        >>> # Complete transaction with metrics
+        >>> result = await metrics_node.execute(
+        ...     operation="end_transaction",
+        ...     transaction_id="txn_12345",
+        ...     status="success",
+        ...     custom_metrics={"items_processed": 25, "db_queries": 3}
+        ... )
+
+        >>> # Get aggregated metrics
+        >>> result = await metrics_node.execute(
+        ...     operation="get_aggregated",
+        ...     metric_names=["order_processing"],
+        ...     aggregation_window=300,  # 5 minutes
+        ...     export_format="prometheus"
+        ... )
+    """
+
+    def __init__(self, **kwargs):
+        """Initialize the transaction metrics node."""
+        super().__init__(**kwargs)
+        self._active_transactions: Dict[str, TransactionMetric] = {}
+        self._completed_transactions: List[TransactionMetric] = []
+        self._metric_buffer = defaultdict(list)
+        self._last_aggregation_time = time.time()
+        self.logger.info(f"Initialized TransactionMetricsNode: {self.id}")
+
+    def get_parameters(self) -> Dict[str, NodeParameter]:
+        """Define the parameters this node accepts."""
+        return {
+            "operation": NodeParameter(
+                name="operation",
+                type=str,
+                required=True,
+                description="Operation to perform (start_transaction, end_transaction, get_metrics, get_aggregated)",
+            ),
+            "transaction_id": NodeParameter(
+                name="transaction_id",
+                type=str,
+                required=False,
+                description="Unique transaction identifier",
+            ),
+            "name": NodeParameter(
+                name="name",
+                type=str,
+                required=False,
+                description="Transaction/metric name",
+            ),
+            "status": NodeParameter(
+                name="status",
+                type=str,
+                required=False,
+                default="success",
+                description="Transaction status (success, error, timeout)",
+            ),
+            "error": NodeParameter(
+                name="error",
+                type=str,
+                required=False,
+                description="Error message if transaction failed",
+            ),
+            "tags": NodeParameter(
+                name="tags",
+                type=dict,
+                required=False,
+                default={},
+                description="Tags for metric grouping and filtering",
+            ),
+            "custom_metrics": NodeParameter(
+                name="custom_metrics",
+                type=dict,
+                required=False,
+                default={},
+                description="Custom metrics to attach to transaction",
+            ),
+            "metric_names": NodeParameter(
+                name="metric_names",
+                type=list,
+                required=False,
+                default=[],
+                description="List of metric names to retrieve",
+            ),
+            "aggregation_window": NodeParameter(
+                name="aggregation_window",
+                type=float,
+                required=False,
+                default=60.0,
+                description="Time window for aggregation in seconds",
+            ),
+            "aggregation_types": NodeParameter(
+                name="aggregation_types",
+                type=list,
+                required=False,
+                default=["count", "avg", "p50", "p95", "p99"],
+                description="Types of aggregation to perform",
+            ),
+            "export_format": NodeParameter(
+                name="export_format",
+                type=str,
+                required=False,
+                default="json",
+                description="Export format (json, prometheus, cloudwatch, datadog, opentelemetry)",
+            ),
+            "include_raw": NodeParameter(
+                name="include_raw",
+                type=bool,
+                required=False,
+                default=False,
+                description="Include raw transaction data in response",
+            ),
+        }
+
+    def get_output_schema(self) -> Dict[str, NodeParameter]:
+        """Define the output schema for this node."""
+        return {
+            "metrics": NodeParameter(
+                name="metrics",
+                type=Any,
+                description="Transaction metrics in requested format",
+            ),
+            "transaction_count": NodeParameter(
+                name="transaction_count",
+                type=int,
+                description="Number of transactions processed",
+            ),
+            "total_transactions": NodeParameter(
+                name="total_transactions",
+                type=int,
+                description="Total number of transactions (alias for transaction_count)",
+            ),
+            "success_rate": NodeParameter(
+                name="success_rate",
+                type=float,
+                description="Success rate of transactions (0.0 to 1.0)",
+            ),
+            "aggregations": NodeParameter(
+                name="aggregations", type=dict, description="Aggregated metric data"
+            ),
+            "export_format": NodeParameter(
+                name="export_format", type=str, description="Format of exported metrics"
+            ),
+            "timestamp": NodeParameter(
+                name="timestamp", type=str, description="ISO timestamp of operation"
+            ),
+            "status": NodeParameter(
+                name="status", type=str, description="Operation status"
+            ),
+        }
+
+    async def async_run(self, **kwargs) -> Dict[str, Any]:
+        """Execute transaction metrics operation."""
+        operation = kwargs.get("operation")
+
+        try:
+            if operation == "start_transaction":
+                return await self._start_transaction(**kwargs)
+            elif operation == "end_transaction":
+                return await self._end_transaction(**kwargs)
+            elif operation == "complete_transaction":
+                return await self._end_transaction(**kwargs)  # Same as end_transaction
+            elif operation == "get_metrics":
+                return await self._get_metrics(**kwargs)
+            elif operation == "get_aggregated":
+                return await self._get_aggregated_metrics(**kwargs)
+            else:
+                raise ValueError(f"Unknown operation: {operation}")
+
+        except Exception as e:
+            self.logger.error(f"Transaction metrics operation failed: {str(e)}")
+            raise NodeExecutionError(f"Failed to process transaction metrics: {str(e)}")
+
+    async def _start_transaction(self, **kwargs) -> Dict[str, Any]:
+        """Start tracking a new transaction."""
+        transaction_id = kwargs.get("transaction_id")
+        if not transaction_id:
+            raise ValueError("transaction_id is required for start_transaction")
+
+        name = kwargs.get("name", "unnamed_transaction")
+        tags = kwargs.get("tags", {})
+
+        # Create new transaction metric
+        metric = TransactionMetric(
+            transaction_id=transaction_id,
+            name=name,
+            start_time=time.time(),
+            tags=tags,
+            status="in_progress",
+        )
+
+        self._active_transactions[transaction_id] = metric
+
+        self.logger.debug(f"Started transaction {transaction_id} ({name})")
+
+        return {
+            "metrics": {"transaction_id": transaction_id, "status": "started"},
+            "transaction_count": 1,
+            "total_transactions": 1,  # Alias for backward compatibility
+            "success_rate": 1.0,  # Starting transaction is optimistically successful
+            "aggregations": {},
+            "export_format": "json",
+            "timestamp": datetime.now(UTC).isoformat(),
+            "status": "success",
+        }
+
+    async def _end_transaction(self, **kwargs) -> Dict[str, Any]:
+        """Complete a transaction and record metrics."""
+        transaction_id = kwargs.get("transaction_id")
+        if not transaction_id:
+            raise ValueError("transaction_id is required for end_transaction")
+
+        if transaction_id not in self._active_transactions:
+            raise ValueError(f"Transaction {transaction_id} not found")
+
+        metric = self._active_transactions.pop(transaction_id)
+
+        # Update transaction metrics
+        metric.end_time = time.time()
+        metric.duration = metric.end_time - metric.start_time
+        metric.status = kwargs.get("status", "success")
+        metric.error = kwargs.get("error")
+        metric.custom_metrics = kwargs.get("custom_metrics", {})
+
+        # Store completed transaction
+        self._completed_transactions.append(metric)
+        self._metric_buffer[metric.name].append(metric)
+
+        # Clean old metrics from buffer (keep last hour)
+        cutoff_time = time.time() - 3600
+        self._completed_transactions = [
+            m for m in self._completed_transactions if m.start_time > cutoff_time
+        ]
+
+        self.logger.debug(
+            f"Completed transaction {transaction_id} ({metric.name}) "
+            f"in {metric.duration:.3f}s with status {metric.status}"
+        )
+
+        return {
+            "metrics": {
+                "transaction_id": transaction_id,
+                "duration": metric.duration,
+                "status": metric.status,
+            },
+            "transaction_count": 1,
+            "total_transactions": 1,  # Alias for backward compatibility
+            "success_rate": (
+                1.0 if metric.status == "success" else 0.0
+            ),  # Based on this transaction
+            "aggregations": {},
+            "export_format": "json",
+            "timestamp": datetime.now(UTC).isoformat(),
+            "status": "success",
+        }
+
+    async def _get_metrics(self, **kwargs) -> Dict[str, Any]:
+        """Get raw transaction metrics."""
+        metric_names = kwargs.get("metric_names", [])
+        include_raw = kwargs.get("include_raw", False)
+        export_format = MetricExportFormat(kwargs.get("export_format", "json"))
+
+        # Filter metrics by name if specified
+        if metric_names:
+            filtered_metrics = [
+                m for m in self._completed_transactions if m.name in metric_names
+            ]
+        else:
+            filtered_metrics = self._completed_transactions
+
+        # Calculate success rate
+        total_metrics = len(filtered_metrics)
+        successful_metrics = len([m for m in filtered_metrics if m.status == "success"])
+        success_rate = successful_metrics / total_metrics if total_metrics > 0 else 1.0
+
+        # Format output
+        if export_format == MetricExportFormat.JSON:
+            if include_raw:
+                metrics_data = [self._serialize_metric(m) for m in filtered_metrics]
+            else:
+                metrics_data = {
+                    "transaction_count": len(filtered_metrics),
+                    "metric_names": list(set(m.name for m in filtered_metrics)),
+                    "success_rate": success_rate,
+                }
+        else:
+            metrics_data = self._format_metrics(filtered_metrics, export_format)
+
+        return {
+            "metrics": metrics_data,
+            "transaction_count": len(filtered_metrics),
+            "total_transactions": len(
+                filtered_metrics
+            ),  # Alias for backward compatibility
+            "success_rate": success_rate,  # Add success rate to top level
+            "aggregations": {},
+            "export_format": export_format.value,
+            "timestamp": datetime.now(UTC).isoformat(),
+            "status": "success",
+        }
+
+    async def _get_aggregated_metrics(self, **kwargs) -> Dict[str, Any]:
+        """Get aggregated transaction metrics."""
+        metric_names = kwargs.get("metric_names", [])
+        aggregation_window = kwargs.get("aggregation_window", 60.0)
+        aggregation_types = kwargs.get(
+            "aggregation_types", ["count", "avg", "p50", "p95", "p99"]
+        )
+        export_format = MetricExportFormat(kwargs.get("export_format", "json"))
+
+        # Calculate time window
+        current_time = time.time()
+        window_start = current_time - aggregation_window
+
+        # Aggregate metrics by name
+        aggregations = {}
+
+        for name, metrics in self._metric_buffer.items():
+            if metric_names and name not in metric_names:
+                continue
+
+            # Filter metrics within window
+            window_metrics = [
+                m
+                for m in metrics
+                if m.start_time >= window_start and m.duration is not None
+            ]
+
+            if not window_metrics:
+                continue
+
+            # Calculate aggregations
+            aggregations[name] = self._calculate_aggregations(
+                window_metrics, aggregation_types
+            )
+
+        # Convert aggregations to JSON-serializable format first
+        serialized_aggregations = {}
+        for name, agg in aggregations.items():
+            serialized_aggregations[name] = {
+                "name": agg.name,
+                "count": agg.count,
+                "sum_duration": agg.sum_duration,
+                "min_duration": agg.min_duration,
+                "max_duration": agg.max_duration,
+                "avg_duration": agg.avg_duration,
+                "percentiles": agg.percentiles,
+                "success_count": agg.success_count,
+                "error_count": agg.error_count,
+                "error_rate": agg.error_rate,
+                "tags": agg.tags,
+            }
+
+        # Format output
+        if export_format == MetricExportFormat.JSON:
+            formatted_metrics = serialized_aggregations
+        else:
+            formatted_metrics = self._format_aggregated_metrics(
+                aggregations, export_format
+            )
+
+        transaction_count = (
+            sum(agg.count for agg in aggregations.values()) if aggregations else 0
+        )
+
+        # Calculate overall success rate from aggregations
+        total_success = (
+            sum(agg.success_count for agg in aggregations.values())
+            if aggregations
+            else 0
+        )
+        success_rate = (
+            total_success / transaction_count if transaction_count > 0 else 1.0
+        )
+
+        return {
+            "metrics": formatted_metrics,
+            "transaction_count": transaction_count,
+            "total_transactions": transaction_count,  # Alias for backward compatibility
+            "success_rate": success_rate,  # Calculated from aggregations
+            "aggregations": serialized_aggregations,
+            "export_format": export_format.value,
+            "timestamp": datetime.now(UTC).isoformat(),
+            "status": "success",
+        }
+
+    def _calculate_aggregations(
+        self, metrics: List[TransactionMetric], aggregation_types: List[str]
+    ) -> AggregatedMetrics:
+        """Calculate aggregated metrics from transaction list."""
+        if not metrics:
+            return None
+
+        durations = [m.duration for m in metrics if m.duration is not None]
+        if not durations:
+            return None
+
+        # Sort durations for percentile calculations
+        sorted_durations = sorted(durations)
+
+        # Basic statistics
+        count = len(metrics)
+        sum_duration = sum(durations)
+        min_duration = min(durations)
+        max_duration = max(durations)
+        avg_duration = sum_duration / count
+
+        # Success/error counts
+        success_count = sum(1 for m in metrics if m.status == "success")
+        error_count = count - success_count
+        error_rate = error_count / count if count > 0 else 0.0
+
+        # Calculate percentiles
+        percentiles = {}
+        percentile_mappings = {
+            "p50": 50,
+            "p75": 75,
+            "p90": 90,
+            "p95": 95,
+            "p99": 99,
+            "p999": 99.9,
+        }
+
+        for agg_type in aggregation_types:
+            if agg_type in percentile_mappings:
+                percentile = percentile_mappings[agg_type]
+                index = int(len(sorted_durations) * (percentile / 100.0))
+                index = min(index, len(sorted_durations) - 1)
+                percentiles[agg_type] = sorted_durations[index]
+
+        # Aggregate tags (use most common values)
+        tag_counts = defaultdict(lambda: defaultdict(int))
+        for metric in metrics:
+            for tag_key, tag_value in metric.tags.items():
+                tag_counts[tag_key][tag_value] += 1
+
+        aggregated_tags = {}
+        for tag_key, value_counts in tag_counts.items():
+            # Use most common tag value
+            most_common = max(value_counts.items(), key=lambda x: x[1])
+            aggregated_tags[tag_key] = most_common[0]
+
+        return AggregatedMetrics(
+            name=metrics[0].name,
+            count=count,
+            sum_duration=sum_duration,
+            min_duration=min_duration,
+            max_duration=max_duration,
+            avg_duration=avg_duration,
+            percentiles=percentiles,
+            success_count=success_count,
+            error_count=error_count,
+            error_rate=error_rate,
+            tags=aggregated_tags,
+        )
+
+    def _serialize_metric(self, metric: TransactionMetric) -> Dict[str, Any]:
+        """Serialize a transaction metric to dictionary."""
+        return {
+            "transaction_id": metric.transaction_id,
+            "name": metric.name,
+            "start_time": metric.start_time,
+            "end_time": metric.end_time,
+            "duration": metric.duration,
+            "status": metric.status,
+            "error": metric.error,
+            "tags": metric.tags,
+            "custom_metrics": metric.custom_metrics,
+        }
+
+    def _format_metrics(
+        self, metrics: List[TransactionMetric], format: MetricExportFormat
+    ) -> Union[str, Dict[str, Any]]:
+        """Format metrics for export."""
+        if format == MetricExportFormat.PROMETHEUS:
+            return self._format_prometheus(metrics)
+        elif format == MetricExportFormat.CLOUDWATCH:
+            return self._format_cloudwatch(metrics)
+        elif format == MetricExportFormat.DATADOG:
+            return self._format_datadog(metrics)
+        elif format == MetricExportFormat.OPENTELEMETRY:
+            return self._format_opentelemetry(metrics)
+        else:
+            return [self._serialize_metric(m) for m in metrics]
+
+    def _format_prometheus(self, metrics: List[TransactionMetric]) -> str:
+        """Format metrics in Prometheus exposition format."""
+        lines = []
+
+        # Group by metric name
+        by_name = defaultdict(list)
+        for m in metrics:
+            by_name[m.name].append(m)
+
+        for name, metric_list in by_name.items():
+            # Transaction duration histogram
+            lines.append("# TYPE transaction_duration_seconds histogram")
+            lines.append(
+                "# HELP transaction_duration_seconds Transaction duration in seconds"
+            )
+
+            for metric in metric_list:
+                if metric.duration is not None:
+                    labels = self._format_prometheus_labels(metric.tags)
+                    lines.append(
+                        f'transaction_duration_seconds{{{labels},name="{name}"}} {metric.duration}'
+                    )
+
+            # Success/error counters
+            success_count = sum(1 for m in metric_list if m.status == "success")
+            error_count = len(metric_list) - success_count
+
+            lines.append("# TYPE transaction_total counter")
+            lines.append("# HELP transaction_total Total number of transactions")
+            lines.append(
+                f'transaction_total{{name="{name}",status="success"}} {success_count}'
+            )
+            lines.append(
+                f'transaction_total{{name="{name}",status="error"}} {error_count}'
+            )
+
+        return "\n".join(lines)
+
+    def _format_prometheus_labels(self, tags: Dict[str, str]) -> str:
+        """Format tags as Prometheus labels."""
+        label_parts = []
+        for k, v in tags.items():
+            # Escape quotes and backslashes
+            v = v.replace("\\", "\\\\").replace('"', '\\"')
+            label_parts.append(f'{k}="{v}"')
+        return ",".join(label_parts)
+
+    def _format_cloudwatch(self, metrics: List[TransactionMetric]) -> Dict[str, Any]:
+        """Format metrics for AWS CloudWatch."""
+        cloudwatch_metrics = []
+
+        for metric in metrics:
+            if metric.duration is not None:
+                cw_metric = {
+                    "MetricName": f"TransactionDuration_{metric.name}",
+                    "Value": metric.duration * 1000,  # Convert to milliseconds
+                    "Unit": "Milliseconds",
+                    "Timestamp": datetime.fromtimestamp(
+                        metric.start_time, UTC
+                    ).isoformat(),
+                    "Dimensions": [
+                        {"Name": k, "Value": v} for k, v in metric.tags.items()
+                    ],
+                }
+                cloudwatch_metrics.append(cw_metric)
+
+                # Add custom metrics
+                for custom_name, custom_value in metric.custom_metrics.items():
+                    cw_custom = {
+                        "MetricName": f"Custom_{metric.name}_{custom_name}",
+                        "Value": custom_value,
+                        "Unit": "Count",
+                        "Timestamp": datetime.fromtimestamp(
+                            metric.start_time, UTC
+                        ).isoformat(),
+                        "Dimensions": [
+                            {"Name": k, "Value": v} for k, v in metric.tags.items()
+                        ],
+                    }
+                    cloudwatch_metrics.append(cw_custom)
+
+        return {"MetricData": cloudwatch_metrics}
+
+    def _format_datadog(self, metrics: List[TransactionMetric]) -> Dict[str, Any]:
+        """Format metrics for DataDog."""
+        series = []
+
+        for metric in metrics:
+            if metric.duration is not None:
+                # Duration metric
+                dd_metric = {
+                    "metric": "transaction.duration",
+                    "points": [[int(metric.start_time), metric.duration]],
+                    "type": "gauge",
+                    "tags": [f"{k}:{v}" for k, v in metric.tags.items()]
+                    + [f"transaction_name:{metric.name}"],
+                }
+                series.append(dd_metric)
+
+                # Status counter
+                status_metric = {
+                    "metric": "transaction.count",
+                    "points": [[int(metric.start_time), 1]],
+                    "type": "count",
+                    "tags": [f"{k}:{v}" for k, v in metric.tags.items()]
+                    + [f"transaction_name:{metric.name}", f"status:{metric.status}"],
+                }
+                series.append(status_metric)
+
+        return {"series": series}
+
+    def _format_opentelemetry(self, metrics: List[TransactionMetric]) -> Dict[str, Any]:
+        """Format metrics in OpenTelemetry format."""
+        otel_metrics = []
+
+        for metric in metrics:
+            if metric.duration is not None:
+                otel_metric = {
+                    "name": "transaction.duration",
+                    "description": f"Duration of {metric.name} transaction",
+                    "unit": "s",
+                    "data": {
+                        "type": "Gauge",
+                        "data_points": [
+                            {
+                                "attributes": {
+                                    **metric.tags,
+                                    "transaction.name": metric.name,
+                                    "transaction.status": metric.status,
+                                },
+                                "time_unix_nano": int(metric.start_time * 1e9),
+                                "value": metric.duration,
+                            }
+                        ],
+                    },
+                }
+                otel_metrics.append(otel_metric)
+
+        return {"resource_metrics": [{"scope_metrics": [{"metrics": otel_metrics}]}]}
+
+    def _format_aggregated_metrics(
+        self, aggregations: Dict[str, AggregatedMetrics], format: MetricExportFormat
+    ) -> Union[str, Dict[str, Any]]:
+        """Format aggregated metrics for export."""
+        if format == MetricExportFormat.PROMETHEUS:
+            lines = []
+
+            for name, agg in aggregations.items():
+                labels = self._format_prometheus_labels(agg.tags)
+                base_labels = f'name="{name}"' + (f",{labels}" if labels else "")
+
+                # Summary metrics
+                lines.append("# TYPE transaction_duration_summary summary")
+                lines.append(
+                    f"transaction_duration_summary_count{{{base_labels}}} {agg.count}"
+                )
+                lines.append(
+                    f"transaction_duration_summary_sum{{{base_labels}}} {agg.sum_duration}"
+                )
+
+                # Percentiles
+                for percentile_name, value in agg.percentiles.items():
+                    quantile = percentile_name[1:]  # Remove 'p' prefix
+                    lines.append(
+                        f'transaction_duration_summary{{{base_labels},quantile="0.{quantile}"}} {value}'
+                    )
+
+                # Error rate
+                lines.append("# TYPE transaction_error_rate gauge")
+                lines.append(
+                    f"transaction_error_rate{{{base_labels}}} {agg.error_rate}"
+                )
+
+            return "\n".join(lines)
+
+        else:
+            # For other formats, return structured data
+            return {
+                name: {
+                    "count": agg.count,
+                    "duration": {
+                        "sum": agg.sum_duration,
+                        "min": agg.min_duration,
+                        "max": agg.max_duration,
+                        "avg": agg.avg_duration,
+                        **agg.percentiles,
+                    },
+                    "success_count": agg.success_count,
+                    "error_count": agg.error_count,
+                    "error_rate": agg.error_rate,
+                    "tags": agg.tags,
+                }
+                for name, agg in aggregations.items()
+            }
+
+    def run(self, **kwargs) -> Dict[str, Any]:
+        """Synchronous wrapper for compatibility."""
+        import asyncio
+
+        return asyncio.run(self.async_run(**kwargs))