feat: improve error handling for log_metrics decorator (#71)

* feat: dont throw exception by default from log_metrics if no metrics are emitted

* docs: update details for change to error handling

* chore: rename parameter for clarity

* fix: correct bug in exception handling from previous commits

add raise_on_empty_metrics param to partial func call
revert mistake in change to test for exception propagation

* improv: change log debug statement to warning when no metrics are present

* docs: add note on suppressing warning for empty metric set

* improv: add warning for manual serialization

Signed-off-by: heitorlessa <[email protected]>

* improv: whitespace and warning supress as info

Signed-off-by: heitorlessa <[email protected]>

Co-authored-by: heitorlessa <[email protected]>

Author: Tom McCarthy

aws_lambda_powertools/metrics/metrics.py

@@ -2,6 +2,7 @@
 import json
 import logging
 import os
+import warnings
 from typing import Any, Callable
 
 from .base import MetricManager
@@ -82,7 +83,12 @@ def clear_metrics(self):
         self.metric_set.clear()
         self.dimension_set.clear()
 
-    def log_metrics(self, lambda_handler: Callable[[Any, Any], Any] = None, capture_cold_start_metric: bool = False):
+    def log_metrics(
+        self,
+        lambda_handler: Callable[[Any, Any], Any] = None,
+        capture_cold_start_metric: bool = False,
+        raise_on_empty_metrics: bool = False,
+    ):
         """Decorator to serialize and publish metrics at the end of a function execution.
 
         Be aware that the log_metrics **does call* the decorated function (e.g. lambda_handler).
@@ -102,6 +108,10 @@ def handler(event, context)
         ----------
         lambda_handler : Callable[[Any, Any], Any], optional
             Lambda function handler, by default None
+        capture_cold_start_metric : bool, optional
+            Captures cold start metric, by default False
+        raise_on_empty_metrics : bool, optional
+            Raise exception if no metrics are emitted, by default False
 
         Raises
         ------
@@ -113,7 +123,11 @@ def handler(event, context)
         # Return a partial function with args filled
         if lambda_handler is None:
             logger.debug("Decorator called with parameters")
-            return functools.partial(self.log_metrics, capture_cold_start_metric=capture_cold_start_metric)
+            return functools.partial(
+                self.log_metrics,
+                capture_cold_start_metric=capture_cold_start_metric,
+                raise_on_empty_metrics=raise_on_empty_metrics,
+            )
 
         @functools.wraps(lambda_handler)
         def decorate(event, context):
@@ -122,10 +136,13 @@ def decorate(event, context):
                 if capture_cold_start_metric:
                     self.__add_cold_start_metric(context=context)
             finally:
-                metrics = self.serialize_metric_set()
-                self.clear_metrics()
-                logger.debug("Publishing metrics", {"metrics": metrics})
-                print(json.dumps(metrics))
+                if not raise_on_empty_metrics and not self.metric_set:
+                    warnings.warn("No metrics to publish, skipping")
+                else:
+                    metrics = self.serialize_metric_set()
+                    self.clear_metrics()
+                    logger.debug("Publishing metrics", {"metrics": metrics})
+                    print(json.dumps(metrics))
 
             return response
 

docs/content/core/metrics.mdx

@@ -107,15 +107,31 @@ def lambda_handler(evt, ctx):
     ...
 ```
 
-`log_metrics` decorator **validates**, **serializes**, and **flushes** all your metrics. During metrics validation, if any of the following criteria is met, `SchemaValidationError` exception will be raised:
+`log_metrics` decorator **validates**, **serializes**, and **flushes** all your metrics. During metrics validation, if no metrics are provided then a warning will be logged, but no exception will be raised.
 
-* At least of one Metric and Dimension
+If metrics are provided, and any of the following criteria are not met, `SchemaValidationError` exception will be raised:
+
+* Minimum of 1 dimension
 * Maximum of 9 dimensions
 * Namespace is set, and no more than one
 * Metric units must be [supported by CloudWatch](https://docs.aws.amazon.com/AmazonCloudWatch/latest/APIReference/API_MetricDatum.html)
 
+If you want to ensure that at least one metric is emitted, you can pass `raise_on_empty_metrics` to the **log_metrics** decorator:
+
+```python:title=lambda_handler.py
+from aws_lambda_powertools.metrics import Metrics
+
+@metrics.log_metrics(raise_on_empty_metrics=True) # highlight-line
+def lambda_handler(evt, ctx):
+    ...
+```
+
+<Note type="info">
+  If you expect your function to execute without publishing metrics every time, you can suppress the warning with <strong><code>warnings.filterwarnings("ignore", "No metrics to publish*")</code></strong>.
+</Note><br/>
+
 <Note type="warning">
-  When nesting multiple middlwares, you should use <strong><code>log_metrics</code> as your last decorator wrapping all subsequent ones</strong>.
+  When nesting multiple middlewares, you should use <strong><code>log_metrics</code> as your last decorator wrapping all subsequent ones</strong>.
 </Note><br/>
 
 ```python:title=lambda_handler_nested_middlewares.py
@@ -138,6 +154,10 @@ def lambda_handler(evt, ctx):
 
 If you prefer not to use `log_metrics` because you might want to encapsulate additional logic when doing so, you can manually flush and clear metrics as follows:
 
+<Note type="warning">
+  Metrics, dimensions and namespace validation still applies.
+</Note><br/>
+
 ```python:title=manual_metric_serialization.py
 import json
 from aws_lambda_powertools import Metrics

tests/functional/test_metrics.py

@@ -1,4 +1,5 @@
 import json
+import warnings
 from collections import namedtuple
 from typing import Any, Dict, List
 
@@ -350,13 +351,13 @@ def test_log_no_metrics_error_propagation(capsys, metric, dimension, namespace):
     # GIVEN Metrics is initialized
     my_metrics = Metrics()
 
-    @my_metrics.log_metrics
+    @my_metrics.log_metrics(raise_on_empty_metrics=True)
     def lambda_handler(evt, context):
-        # WHEN log_metrics is used despite having no metrics
+        # WHEN log_metrics is used with raise_on_empty_metrics param and has no metrics
         # and the function decorated also raised an exception
         raise ValueError("Bubble up")
 
-    # THEN we should first raise SchemaValidationError as the main exception
+    # THEN the raised exception should be
     with pytest.raises(SchemaValidationError):
         lambda_handler({}, {})
 
@@ -632,3 +633,19 @@ def lambda_handler(evt, context):
     assert "ColdStart" not in output
 
     assert "function_name" not in output
+
+
+def test_log_metrics_decorator_no_metrics(dimensions, namespace):
+    # GIVEN Metrics is initialized
+    my_metrics = Metrics(namespace=namespace["name"], service="test_service")
+
+    # WHEN using the log_metrics decorator and no metrics have been added
+    @my_metrics.log_metrics
+    def lambda_handler(evt, context):
+        pass
+
+    # THEN it should raise a warning instead of throwing an exception
+    with warnings.catch_warnings(record=True) as w:
+        lambda_handler({}, {})
+        assert len(w) == 1
+        assert str(w[-1].message) == "No metrics to publish, skipping"