Merge branch 'main' of github.com:open-telemetry/opentelemetry-python into codeboten/generate-config-model-from-schema

Author: MikeGoldsmith

.github/workflows/benchmarks.yml

@@ -42,3 +42,4 @@ jobs:
         # Alert with a commit comment on possible performance regression
         alert-threshold: '200%'
         comment-on-alert: true
+        alert-comment-cc-users: "@open-telemetry/python-approvers,@open-telemetry/python-maintainers"

CHANGELOG.md

@@ -12,14 +12,26 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## Unreleased
 
+- `opentelemetry-exporter-otlp-proto-grpc`: Fix re-initialization of gRPC channel on UNAVAILABLE error
+  ([#4825](https://github.com/open-telemetry/opentelemetry-python/pull/4825))
+- `opentelemetry-exporter-prometheus`: Fix duplicate HELP/TYPE declarations for metrics with different label sets
+  ([#4868](https://github.com/open-telemetry/opentelemetry-python/issues/4868))
+- Allow loading all resource detectors by setting `OTEL_EXPERIMENTAL_RESOURCE_DETECTORS` to `*`
+  ([#4819](https://github.com/open-telemetry/opentelemetry-python/pull/4819))
 - `opentelemetry-sdk`: Fix the type hint of the `_metrics_data` property to allow `None`
-  ([#4837](https://github.com/open-telemetry/opentelemetry-python/pull/4837)
+  ([#4837](https://github.com/open-telemetry/opentelemetry-python/pull/4837)).
 - Regenerate opentelemetry-proto code with v1.9.0 release
   ([#4840](https://github.com/open-telemetry/opentelemetry-python/pull/4840))
 - Add python 3.14 support
   ([#4798](https://github.com/open-telemetry/opentelemetry-python/pull/4798))
 - Silence events API warnings for internal users
   ([#4847](https://github.com/open-telemetry/opentelemetry-python/pull/4847))
+- Prevent possible endless recursion from happening in `SimpleLogRecordProcessor.on_emit`,
+  ([#4799](https://github.com/open-telemetry/opentelemetry-python/pull/4799)) and ([#4867](https://github.com/open-telemetry/opentelemetry-python/pull/4867)).
+- Make ConcurrentMultiSpanProcessor fork safe
+  ([#4862](https://github.com/open-telemetry/opentelemetry-python/pull/4862))
+- `opentelemetry-exporter-otlp-proto-http`: fix retry logic and error handling for connection failures in trace, metric, and log exporters
+  ([#4709](https://github.com/open-telemetry/opentelemetry-python/pull/4709))
 
 ## Version 1.39.0/0.60b0 (2025-12-03)
 
@@ -87,7 +99,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ([#4654](https://github.com/open-telemetry/opentelemetry-python/pull/4654)).
 - Fix type checking for built-in metric exporters
   ([#4820](https://github.com/open-telemetry/opentelemetry-python/pull/4820))
-  
+
 ## Version 1.38.0/0.59b0 (2025-10-16)
 
 - Add `rstcheck` to pre-commit to stop introducing invalid RST

README.md

@@ -111,7 +111,7 @@ For more information about the maintainer role, see the [community repository](h
 ### Approvers
 
 - [Dylan Russell](https://github.com/dylanrussell), Google
-- [Emídio Neto](https://github.com/emdneto), PicPay
+- [Emídio Neto](https://github.com/emdneto), Independent
 - [Héctor Hernández](https://github.com/hectorhdzg), Microsoft
 - [Jeremy Voss](https://github.com/jeremydvoss), Microsoft
 - [Liudmila Molkova](https://github.com/lmolkova), Grafana Labs

RELEASING.md

@@ -33,6 +33,18 @@
   * Review and merge the pull request that it creates for updating the version.
 * Note: If you are doing a patch release in `-core` repo, you should also do an equivalent patch release in `-contrib` repo (even if there's no fix to release), otherwise tests in CI will fail.
 
+### Note on `contrib.yml` Workflow Behavior
+
+The [contrib.yml](https://github.com/open-telemetry/opentelemetry-python/blob/main/.github/workflows/contrib.yml) workflow in the core repository references reusable workflows from opentelemetry-python-contrib using the hard-coded `main` branch.
+
+Because `uses:` statements cannot receive environment variables and workflows cannot patch or modify other workflows, this reference cannot dynamically follow release branches as we are doing in other workflows.
+
+As a result, when preparing a release branch that contains a different set of instrumentations (e.g., older branches without newly added tox environments), CI may attempt to run tests that do not exist on tox in that branch. In this case:
+
+* It is safe to merge the release PR even if the contrib workflow fails for this reason, or
+
+* Optionally update the contrib.yml workflow to point to the corresponding release branch before running CI.
+
 ## Making the release
 
 * Run the [Release workflow](https://github.com/open-telemetry/opentelemetry-python/actions/workflows/release.yml).

docs/examples/sqlcommenter/README.rst

@@ -5,7 +5,7 @@ This is an example of how to use OpenTelemetry Python instrumention with
 sqlcommenter to enrich database query statements with contextual information.
 For more information on sqlcommenter concepts, see:
 
-* `Semantic Conventions - Database Spans <https://github.com/open-telemetry/semantic-conventions/blob/main/docs/database/database-spans.md#sql-commenter>`_
+* `Semantic Conventions - Database Spans <https://github.com/open-telemetry/semantic-conventions/blob/main/docs/db/database-spans.md#sql-commenter>`_
 * `sqlcommenter <https://google.github.io/sqlcommenter/>`_
 
 The source files of this example are available `here <https://github.com/open-telemetry/opentelemetry-python/tree/main/docs/examples/sqlcommenter/>`_.
@@ -120,5 +120,5 @@ References
 * `OpenTelemetry Project <https://opentelemetry.io/>`_
 * `OpenTelemetry Collector <https://github.com/open-telemetry/opentelemetry-collector>`_
 * `OpenTelemetry MySQL instrumentation <https://github.com/open-telemetry/opentelemetry-python-contrib/tree/main/instrumentation/opentelemetry-instrumentation-mysql>`_
-* `Semantic Conventions - Database Spans <https://github.com/open-telemetry/semantic-conventions/blob/main/docs/database/database-spans.md#sql-commenter>`_
+* `Semantic Conventions - Database Spans <https://github.com/open-telemetry/semantic-conventions/blob/main/docs/db/database-spans.md#sql-commenter>`_
 * `sqlcommenter <https://google.github.io/sqlcommenter/>`_

exporter/opentelemetry-exporter-otlp-proto-grpc/src/opentelemetry/exporter/otlp/proto/grpc/exporter.py

@@ -12,7 +12,13 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-"""OTLP Exporter"""
+"""OTLP Exporter
+
+This module provides a mixin class for OTLP exporters that send telemetry data
+to an OTLP-compatible receiver via gRPC. It includes a configurable reconnection
+logic to handle transient collector outages.
+
+"""
 
 import random
 import threading
@@ -251,20 +257,27 @@ def _get_credentials(
     if certificate_file:
         client_key_file = environ.get(client_key_file_env_key)
         client_certificate_file = environ.get(client_certificate_file_env_key)
-        return _load_credentials(
+        credentials = _load_credentials(
             certificate_file, client_key_file, client_certificate_file
         )
+        if credentials is not None:
+            return credentials
     return ssl_channel_credentials()
 
 
 # pylint: disable=no-member
 class OTLPExporterMixin(
     ABC, Generic[SDKDataT, ExportServiceRequestT, ExportResultT, ExportStubT]
 ):
-    """OTLP span exporter
+    """OTLP gRPC exporter mixin.
+
+    This class provides the base functionality for OTLP exporters that send
+    telemetry data (spans or metrics) to an OTLP-compatible receiver via gRPC.
+    It includes a configurable reconnection mechanism to handle transient
+    receiver outages.
 
     Args:
-        endpoint: OpenTelemetry Collector receiver endpoint
+        endpoint: OTLP-compatible receiver endpoint
         insecure: Connection type
         credentials: ChannelCredentials object for server authentication
         headers: Headers to send when exporting
@@ -308,6 +321,8 @@ def __init__(
         if parsed_url.netloc:
             self._endpoint = parsed_url.netloc
 
+        self._insecure = insecure
+        self._credentials = credentials
         self._headers = headers or environ.get(OTEL_EXPORTER_OTLP_HEADERS)
         if isinstance(self._headers, str):
             temp_headers = parse_env_headers(self._headers, liberal=True)
@@ -336,37 +351,52 @@ def __init__(
         )
         self._collector_kwargs = None
 
-        compression = (
+        self._compression = (
             environ_to_compression(OTEL_EXPORTER_OTLP_COMPRESSION)
             if compression is None
             else compression
         ) or Compression.NoCompression
 
-        if insecure:
-            self._channel = insecure_channel(
-                self._endpoint,
-                compression=compression,
-                options=self._channel_options,
-            )
-        else:
+        self._channel = None
+        self._client = None
+
+        self._shutdown_in_progress = threading.Event()
+        self._shutdown = False
+
+        if not self._insecure:
             self._credentials = _get_credentials(
-                credentials,
+                self._credentials,
                 _OTEL_PYTHON_EXPORTER_OTLP_GRPC_CREDENTIAL_PROVIDER,
                 OTEL_EXPORTER_OTLP_CERTIFICATE,
                 OTEL_EXPORTER_OTLP_CLIENT_KEY,
                 OTEL_EXPORTER_OTLP_CLIENT_CERTIFICATE,
             )
+
+        self._initialize_channel_and_stub()
+
+    def _initialize_channel_and_stub(self):
+        """
+        Create a new gRPC channel and stub.
+
+        This method is used during initialization and by the reconnection
+        mechanism to reinitialize the channel on transient errors.
+        """
+        if self._insecure:
+            self._channel = insecure_channel(
+                self._endpoint,
+                compression=self._compression,
+                options=self._channel_options,
+            )
+        else:
+            assert self._credentials is not None
             self._channel = secure_channel(
                 self._endpoint,
                 self._credentials,
-                compression=compression,
+                compression=self._compression,
                 options=self._channel_options,
             )
         self._client = self._stub(self._channel)  # type: ignore [reportCallIssue]
 
-        self._shutdown_in_progress = threading.Event()
-        self._shutdown = False
-
     @abstractmethod
     def _translate_data(
         self,
@@ -388,6 +418,8 @@ def _export(
         deadline_sec = time() + self._timeout
         for retry_num in range(_MAX_RETRYS):
             try:
+                if self._client is None:
+                    return self._result.FAILURE
                 self._client.Export(
                     request=self._translate_data(data),
                     metadata=self._headers,
@@ -407,6 +439,26 @@ def _export(
                         retry_info.retry_delay.seconds
                         + retry_info.retry_delay.nanos / 1.0e9
                     )
+
+                # For UNAVAILABLE errors, reinitialize the channel to force reconnection
+                if error.code() == StatusCode.UNAVAILABLE and retry_num == 0:  # type: ignore
+                    logger.debug(
+                        "Reinitializing gRPC channel for %s exporter due to UNAVAILABLE error",
+                        self._exporting,
+                    )
+                    try:
+                        if self._channel:
+                            self._channel.close()
+                    except Exception as e:
+                        logger.debug(
+                            "Error closing channel for %s exporter to %s: %s",
+                            self._exporting,
+                            self._endpoint,
+                            str(e),
+                        )
+                    # Enable channel reconnection for subsequent calls
+                    self._initialize_channel_and_stub()
+
                 if (
                     error.code() not in _RETRYABLE_ERROR_CODES  # type: ignore [reportAttributeAccessIssue]
                     or retry_num + 1 == _MAX_RETRYS
@@ -436,12 +488,19 @@ def _export(
         return self._result.FAILURE  # type: ignore [reportReturnType]
 
     def shutdown(self, timeout_millis: float = 30_000, **kwargs) -> None:
+        """
+        Shut down the exporter.
+
+        Args:
+            timeout_millis: Timeout in milliseconds for shutting down the exporter.
+        """
         if self._shutdown:
             logger.warning("Exporter already shutdown, ignoring call")
             return
         self._shutdown = True
         self._shutdown_in_progress.set()
-        self._channel.close()
+        if self._channel:
+            self._channel.close()
 
     @property
     @abstractmethod

exporter/opentelemetry-exporter-otlp-proto-grpc/tests/test_otlp_exporter_mixin.py

@@ -24,6 +24,7 @@
 from unittest import TestCase
 from unittest.mock import Mock, patch
 
+import grpc
 from google.protobuf.duration_pb2 import (  # pylint: disable=no-name-in-module
     Duration,
 )
@@ -91,8 +92,8 @@ def export(self, spans: Sequence[ReadableSpan]) -> SpanExportResult:
     def _exporting(self):
         return "traces"
 
-    def shutdown(self, timeout_millis=30_000):
-        return OTLPExporterMixin.shutdown(self, timeout_millis)
+    def shutdown(self, timeout_millis: float = 30_000, **kwargs):
+        return OTLPExporterMixin.shutdown(self, timeout_millis, **kwargs)
 
 
 class TraceServiceServicerWithExportParams(TraceServiceServicer):
@@ -513,6 +514,16 @@ def test_timeout_set_correctly(self):
             self.assertEqual(mock_trace_service.num_requests, 2)
             self.assertAlmostEqual(after - before, 1.4, 1)
 
+    def test_channel_options_set_correctly(self):
+        """Test that gRPC channel options are set correctly for keepalive and reconnection"""
+        # This test verifies that the channel is created with the right options
+        # We patch grpc.insecure_channel to ensure it is called without errors
+        with patch(
+            "opentelemetry.exporter.otlp.proto.grpc.exporter.insecure_channel"
+        ) as mock_channel:
+            OTLPSpanExporterForTesting(insecure=True)
+            self.assertTrue(mock_channel.called)
+
     def test_otlp_headers_from_env(self):
         # pylint: disable=protected-access
         # This ensures that there is no other header than standard user-agent.
@@ -536,3 +547,27 @@ def test_permanent_failure(self):
                 warning.records[-1].message,
                 "Failed to export traces to localhost:4317, error code: StatusCode.ALREADY_EXISTS",
             )
+
+    def test_unavailable_reconnects(self):
+        """Test that the exporter reconnects on UNAVAILABLE error"""
+        add_TraceServiceServicer_to_server(
+            TraceServiceServicerWithExportParams(StatusCode.UNAVAILABLE),
+            self.server,
+        )
+
+        # Spy on grpc.insecure_channel to verify it's called for reconnection
+        with patch(
+            "opentelemetry.exporter.otlp.proto.grpc.exporter.insecure_channel",
+            side_effect=grpc.insecure_channel,
+        ) as mock_insecure_channel:
+            # Mock sleep to avoid waiting
+            with patch("time.sleep"):
+                # We expect FAILURE because the server keeps returning UNAVAILABLE
+                # but we want to verify reconnection attempts happened
+                self.exporter.export([self.span])
+
+        # Verify that we attempted to reinitialize the channel (called insecure_channel)
+        # Since the initial channel was created in setUp (unpatched), this call
+        # must be from the reconnection logic.
+        self.assertTrue(mock_insecure_channel.called)
+        # Verify that reconnection enabled flag is set

exporter/opentelemetry-exporter-otlp-proto-http/src/opentelemetry/exporter/otlp/proto/http/_log_exporter/__init__.py

@@ -186,26 +186,42 @@ def export(
         serialized_data = encode_logs(batch).SerializeToString()
         deadline_sec = time() + self._timeout
         for retry_num in range(_MAX_RETRYS):
-            resp = self._export(serialized_data, deadline_sec - time())
-            if resp.ok:
-                return LogRecordExportResult.SUCCESS
             # multiplying by a random number between .8 and 1.2 introduces a +/20% jitter to each backoff.
             backoff_seconds = 2**retry_num * random.uniform(0.8, 1.2)
+            try:
+                resp = self._export(serialized_data, deadline_sec - time())
+                if resp.ok:
+                    return LogRecordExportResult.SUCCESS
+            except requests.exceptions.RequestException as error:
+                reason = error
+                retryable = isinstance(error, ConnectionError)
+                status_code = None
+            else:
+                reason = resp.reason
+                retryable = _is_retryable(resp)
+                status_code = resp.status_code
+
+            if not retryable:
+                _logger.error(
+                    "Failed to export logs batch code: %s, reason: %s",
+                    status_code,
+                    reason,
+                )
+                return LogRecordExportResult.FAILURE
+
             if (
-                not _is_retryable(resp)
-                or retry_num + 1 == _MAX_RETRYS
+                retry_num + 1 == _MAX_RETRYS
                 or backoff_seconds > (deadline_sec - time())
                 or self._shutdown
             ):
                 _logger.error(
-                    "Failed to export logs batch code: %s, reason: %s",
-                    resp.status_code,
-                    resp.text,
+                    "Failed to export logs batch due to timeout, "
+                    "max retries or shutdown."
                 )
                 return LogRecordExportResult.FAILURE
             _logger.warning(
                 "Transient error %s encountered while exporting logs batch, retrying in %.2fs.",
-                resp.reason,
+                reason,
                 backoff_seconds,
             )
             shutdown = self._shutdown_is_occuring.wait(backoff_seconds)