v0.142.0

Compare Source

🛑 Breaking changes 🛑

• all: It's recommended to change the field type in your component configuration to be configoptional.Optional[exporterhelper.QueueBatchConfig] to keep the enabled subfield. Use configoptional.Some(exporterhelper.NewDefaultQueueConfig()) to enable by default. Use configoptional.Default(exporterhelper.NewDefaultQueueConfig()) to disable by default. (#​44320)

• exporter/datadog: Update the Datadog exporter to support the Orchestrator Explorer by accepting receiver/k8sobjects logs and sending Kubernetes data to the Datadog endpoint. (#​44523)
  The Cluster name field in Orchestrator Explorer Configuration has been removed. Use the k8s.cluster.name attribute instead.

• exporter/prometheusremotewrite: Updated to Remote Write 2.0 spec rc.4, requiring Prometheus 3.8.0 or later as the receiving endpoint. (#​44861)
  The upstream Prometheus library updated the Remote Write 2.0 protocol from rc.3 to rc.4 in
  prometheus/prometheus#17411. This renamed CreatedTimestamp to StartTimestamp and moved it
  from the TimeSeries message to individual Sample and Histogram messages. This is a
  wire-protocol incompatibility, so if you are exporting to a Prometheus server, you must upgrade
  it to version 3.8.0 or later to receive data from this exporter.

• extension/googlecloudlogentry_encoding: Parse External Application Load Balancer logs into log record attributes instead of placing it in the body as is. (#​44438)

• pkg/stanza: Allow max_batch_size of 0 for unlimited batching in recombine operator (#​43982)
  The recombine operator now supports setting max_batch_size: 0 to disable batch size limits.
  This allows unlimited batching, letting entries be combined based only on max_log_size and matching conditions.
  If you have max_batch_size: 0 in your config and want to keep the behavior unchanged, change the configuration to max_batch_size: 1.

• processor/cumulativetodelta: Change default max_staleness from 0 (infinite) to 1 hour (#​44427)
  The processor now defaults to a max_staleness of 1 hour instead of 0 (infinite retention).
  This prevents unbounded memory growth in long-running collector instances, especially when tracking metrics with high cardinality or frequently changing attribute values.
  To restore the previous behavior of infinite retention, explicitly set max_staleness: 0 in your configuration.

• processor/resourcedetection: Promote processor.resourcedetection.propagateerrors feature gate to beta (#​44609)

• processor/resourcedetection: Remove deprecated attributes configuration option (#​44610)

• receiver/awss3: Remove the s3_partition config option in favor of s3_partition_format and s3_partition_timezone options. This aligns the S3 receiver more closely with the S3 Exporter. Also add the ability to include or exclude the telemetry type from the file prefix using the file_prefix_include_telemetry_type option. (#​43720)

• receiver/docker_stats: Upgrades default Docker API version to 1.44 to be compatible with recent Docker Engine versions. (#​44279)
  Users requiring an older Docker API version can set the api_version in the docker stats receiver config. The minimum supported API level is not changed, only default.

• receiver/filelog: Move filelog.decompressFingerprint to stable stage (#​44570)

• receiver/prometheus: Promote the receiver.prometheusreceiver.RemoveStartTimeAdjustment feature gate to stable and remove in-receiver metric start time adjustment in favor of the metricstarttime processor, including disabling the created-metric feature gate. (#​44180)
  Previously, users could disable the RemoveStartTimeAdjustment feature gate to temporarily keep the legacy start time adjustment behavior in the Prometheus receiver.
  With this promotion to stable and bounded registration, that gate can no longer be disabled; the receiver will no longer set StartTime on metrics based on process_start_time_seconds, and users should migrate to the metricstarttime processor for equivalent functionality.
  This change also disables the receiver.prometheusreceiver.UseCreatedMetric feature gate, which previously used the <metric>_created series to derive start timestamps for counters, summaries, and histograms when scraping non OpenMetrics protocols.
  However, this does not mean that the _created series is always ignored: when using the OpenMetrics 1.0 protocol, Prometheus itself continues to interpret the _created series as the start timestamp, so only the receiver-side handling for other scrape protocols has been removed.

• receiver/prometheus: Native histogram scraping and ingestion is now controlled by the scrape configuration option scrape_native_histograms. (#​44861)
  The feature gate receiver.prometheusreceiver.EnableNativeHistograms is now stable and enabled by default.
  Native histograms scraped from Prometheus will automatically be converted to OpenTelemetry exponential histograms.

  To enable scraping of native histograms, you must configure scrape_native_histograms: true in your Prometheus
  scrape configuration (either globally or per-job). Additionally, the protobuf scrape protocol must be enabled
  by setting scrape_protocols to include PrometheusProto.

• receiver/prometheusremotewrite: Updated to Remote Write 2.0 spec rc.4, requiring Prometheus 3.8.0 or later (#​44861)
  The upstream Prometheus library updated the Remote Write 2.0 protocol from rc.3 to rc.4 in
  prometheus/prometheus#17411. This renamed CreatedTimestamp to StartTimestamp and moved it
  from the TimeSeries message to individual Sample and Histogram messages. This is a
  wire-protocol incompatibility, so Prometheus versions 3.7.x and earlier will no longer work
  correctly with this receiver. Please upgrade to Prometheus 3.8.0 or later.

🚩 Deprecations 🚩

• processor/k8sattributes: Removes stable k8sattr.fieldExtractConfigRegex.disallow feature gate (#​44694)

• receiver/kafka: Deprecate default_fetch_size parameter for franz-go client (#​43104)
  The default_fetch_size parameter is now deprecated for the franz-go Kafka client and will only be used with the legacy Sarama client.
  Users should configure max_fetch_size instead when using franz-go.
  This deprecation is marked as of v0.142.0.

• receiver/kafka: Support configuring a list of topics and exclude_topics; deprecate topic and exclude_topic (#​44477)

• receiver/prometheus: Deprecate use_start_time_metric and start_time_metric_regex config in favor of the processor metricstarttime (#​44180)

🚀 New components 🚀

• receiver/yanggrpc: Implement the YANG/gRPC receiver (#​43840)

💡 Enhancements 💡

• exporter/elasticsearch: add dynamic data stream routing for connectors (#​44525)

• exporter/kafka: Adds server.address attribute to all Kafka exporter metrics. (#​44649)

• exporter/prometheusremotewrite: Add option to remove service.name, service.instance.id, service.namespace ResourceAttribute from exported metrics (#​44567)

• exporter/signalfx: Support setting default properties for dimension updates to be set lazily as part of configuration (#​44891)

• extension/azure_encoding: Implement general Azure Resource Log parsing functionality (#​41725)

• extension/datadog: Datadog Extension users may view and manage OTel Collectors in Fleet Automation. (#​44666)
  Interested users should read the post on the Datadog Monitor blog and fill out the preview intake form listed there.

• extension/datadog: Adds deployment_type configuration option to the Datadog Extension. (#​44430)
  Users may specify the deployment type of the collector in Datadog Extension configuration to view in Datadog app.
  If the collector is deployed as a gateway (i.e. receiving pipeline telemetry from multiple hosts/sources),
  user should specify "gateway" as the deployment type.
  If the collector is deployed as a daemonset/agent, user should specify "daemonset" as the deployment type.
  The default setting is "unknown" if not set.

• extension/datadog: Adds standard (non-billed) liveness metric otel.datadog_extension.running to ensure host data is shown in Datadog app. (#​44285)

• extension/googlecloudlogentry_encoding: Add support for GCP VPC Flow Log fields for MIG (Managed Instance Group) and Google Service logs. (#​44220)
  Adds support for the following GCP VPC Flow Log fields:

  • Add support for gcp.vpc.flow.{source,destination}.google_service.{type,name,connectivity}
  • Add support for gcp.vpc.flow.{source,destination}.instance.managed_instance_group.{name,region,zone}

• extension/health_check: Added extension.healthcheck.useComponentStatus feature gate to enable v2 component status reporting in healthcheckextension while maintaining backward compatibility by default. (#​42256)

• pkg/ottl: Accept string trace/span/profile IDs for TraceID(), SpanID(), and ProfileID() in OTTL. (#​43429)
  This change allows for a more straightforward use of string values to set trace, span, and profile IDs in OTTL.

• pkg/stanza: New featuregate filelog.windows.caseInsensitive introduced. It will make glob matching is case-insensitive on Windows. (#​40685)
  Previously, any include pattern that included some manner of wildcard (* or **) would
  be case-sensitive on Windows, but Windows filepaths are by default case-insensitive. This meant
  that in a directory with the files a.log and b.LOG, the pattern *.log would previously only
  match a.log. With the filelog.windows.caseInsensitive featuregate enabled, it will match both a.log
  and b.LOG when on Windows. The behaviour is the same as always on other operating systems, as all other
  currently supported platforms for the Collector have case-sensitive filesystems.

• pkg/translator/azurelogs: Added support for Activity Logs Recommendation category (#​43220)

• processor/k8sattributes: Updates semconv version to v1.37.0 (#​44696)

• processor/resourcedetection: Add support for dynamic refresh resource attributes with refresh_interval parameter (#​42663)

• processor/resourcedetection: Update semconv dependency to 1.37.0 which updates the schema url in the data, but no other impact is expected. (#​44726)

• processor/transform: New Transform Processor function set_semconv_span_name() to overwrite the span name with the semantic conventions for HTTP, RPC, messaging, and database spans. (#​43124)
  In other cases, the original span.name is unchanged.
  The primary use of set_semconv_span_name() is alongside the
  Span Metrics Connector
  to address high-cardinality issues in span metrics when span.name does not comply with the OTel requirement
  that span names be low cardinality.

• receiver/azuremonitor: Scrape storage account sub types (#​37220)

• receiver/datadog: Improve the compliance with OTel Semantic Conventions for RPC spans in the Datadog receiver. (#​44671)
  Compliance improvements on spans received via the Datadog receiver when applicable:

  • Set span name according to RPC conventions.
  • Set rpc.method and rpc.service attributes.

• receiver/datadog: Bump OTel Semantic Conventions from v1.30.0 to v1.37.0 in the Datadog receiver. (#​44671)
  Changes in OTel Semantic Conventions v1.37.0 affecting the Datadog receiver:

  • The Datadog tag runtime is now mapped to the OTel attribute container.runtime.name instead of container.runtime.

• receiver/datadog: Improve the compliance with OTel Semantic Conventions for AWS SDK spans in the Datadog receiver. (#​44723)
  Compliance improvements on spans received via the Datadog receiver when applicable:

  • Set span name according to AWS SDK conventions.
  • Set rpc.system, rpc.method and rpc.service attributes.

• receiver/datadog: Add receiver.datadogreceiver.EnableMultiTagParsing feature gate (#​44747)
  The feature flag changes the logic that converts Datadog tags to OpenTelemetry attributes.
  When the flag is enabled, data points that have multiple tags starting with the same key: prefix
  will be turned into an attribute slice (instead of a string) containing all the suffix values.

• receiver/datadog: Improve the compliance with OTel Semantic Conventions for HTTP spans in the Datadog receiver. (#​44722)
  Compliance improvements on spans received via the Datadog receiver when applicable:
  Set span name according to HTTP conventions for web.request and http.request spans.

• receiver/github: Add concurrency limiting to reduce likelihood of hitting secondary rate limits (#​43388)
  Adds concurrency_limit configuration parameter (default: 50) to limit
  concurrent repository processing goroutines. This reduces the likelihood of
  getting 502/504 errors when scraping organizations with >100 repositories.

• receiver/googlecloudpubsub: Exponential backoff streaming restarts (#​44741)

• receiver/kafka: Make session_timeout, heartbeat_interval, max_partition_fetch_size, and max_fetch_wait unconditional in franz-go consumer (#​44839)

• receiver/kafka: Validate that exclude_topics entries in kafkareceiver config are non-empty. (#​44920)

• receiver/oracledb: Added independent collection interval config for Oracle top query metrics collection (#​44607)

• receiver/prometheusremotewrite: Map.PutStr causes excessive memory allocations due to repeated slice expansions (#​44612)

• receiver/splunk_hec: Support parsing JSON array payloads in Splunk HEC receiver (#​43941)

• receiver/sshcheck: Promote sshcheck receiver to beta stability (#​41573)

• receiver/yanggrpc: Promote to alpha stability (#​44783)

🧰 Bug fixes 🧰

• exporter/elasticsearch: Fix hostname mapping in Elasticsearch exporter (#​44874)

  • The exporter now supports to map an otel field to an ecs field only if the ecs field is not already present. This is applied to host.hostname mapping.

• processor/cumulativetodelta: Check whether bucket bounds are the same when verifying whether histograms are comparable (#​44793)

• processor/cumulativetodelta: Fix logic handling ZeroThreshold increases for exponential histograms (#​44793)

• processor/filter: Fix context initialization for metric/datapoint context (#​44813)

• processor/k8sattributes: Fix k8sattr.labelsAnnotationsSingular.allow feature gate to affect config default tag names in addition to runtime extraction (#​39774)

• processor/tail_sampling: Fix a memory leak introduced in 0.141.0 of the tail sampling processor when not blocking on overflow. (#​44884)

• receiver/datadog: The db.instance tag of Datadog database client spans should be mapped to the OTel attribute db.namespace, not to db.collection.name. (#​44702)
  Compliance improvements on spans received via the Datadog receiver when applicable:

  • The db.instance tag is now mapped to the OTel attribute db.namespace instead of db.collection.name.
  • The db.sql.table tag is mapped to the OTel attribute db.collection.name.
  • The db.statement tag is mapped to the OTel attribute db.query.text.

• receiver/datadog: Fix Datadog trace span counting so otelcol_receiver_accepted_spans is not under-reported (#​44865)
  Previously only the last payload's spans were counted, so the otelcol_receiver_accepted_spans metric could be lower than otelcol_exporter_sent_spans in pipelines where they should match.

• receiver/github: Adds corrections to span kind for GitHub events when they are tasks. (#​44667)

• receiver/googlecloudpubsub: Acknowledge messages at restart (#​44706)
  Rewrote the control flow loop so the acknowledgment of messages is more reliable. At stream restart, the messages
  ackIds are resent immediately without an explicit acknowledgment. Outstanding ackIds are only cleared when the
  acknowledgment is sent successfully.

• receiver/googlecloudspanner: Fixed goroutine leaks in ttlcache lifecycle management and applied modernize linter fixes across multiple receivers. (#​44779)

  • Simplified cache lifecycle management by removing unnecessary WaitGroup complexity
  • Added goleak ignores for ttlcache goroutines that don't stop immediately after Stop()

• receiver/kafka: Use max_fetch_size instead of default_fetch_size in franz-go client (#​43104)
  The franz-go Kafka consumer was incorrectly using default_fetch_size (a Sarama-specific setting) instead of max_fetch_size when configuring kgo.FetchMaxBytes.
  This fix ensures the correct parameter is used and adds validation to prevent max_fetch_size from being less than min_fetch_size.
  The default value for max_fetch_size has been changed from 0 (unlimited) to 1048576 (1 MiB) to maintain backward compatibility with the previous (incorrect) behavior.

• receiver/prometheus: Fix HTTP response body leak in target allocator when fetching scrape configs fails (#​44921)
  The getScrapeConfigsResponse function did not close resp.Body on error paths.
  If io.ReadAll or yaml.Unmarshal failed, the response body would leak,
  potentially causing HTTP connection exhaustion.

• receiver/prometheus: Fixes yaml marshaling of prometheus/common/config.Secret types (#​44445)

v0.141.0

Compare Source

🛑 Breaking changes 🛑

• all: fix pprofile DurationNano to be a TypeUint64 (#​44397)

• cmd/otelcontribcol: Removing unmaintained component exporter/carbon (#​38913)

• connector/spanmetrics: Add a feature gate to use the latest semantic conventions for the status code attribute on generated metrics. | This feature gate will replace the status.code attribute on the generated RED metrics with otel.status_code. | It will also replace the values STATUS_CODE_ERROR and STATUS_CODE_OK with ERROR and OK to align with the latest conventions. (#​42103)
  This change is made to align with the latest semantic conventions. | The feature gate is disabled by default, but can be enabled with --feature-gates spanmetrics.statusCodeConvention.useOtelPrefix | or explicitly disabled with --feature-gates -spanmetrics.statusCodeConvention.useOtelPrefix.

• exporter/clickhouse: Add EventName column to logs table (#​42584)
  This column is optional for existing deployments. See project README for notes on how to upgrade your logs table.

• exporter/clickhouse: Add columns for tracking JSON paths in logs + traces (#​43109)
  The JSON columns now include a helper column for keeping track of what keys are inside of the JSON object.
  This change also introduces schema detection logic to reduce breaking changes whenever a column is added.
  Existing users can enable these features by manually adding all the new columns to their table.

• exporter/kafka: exporter.kafkaexporter.UseFranzGo feature gate moved to Stable and is now always enabled (#​44565)
  The franz-go client is now the default and only Kafka client library for the Kafka exporter.
  The feature gate exporter.kafkaexporter.UseFranzGo has been promoted to Stable status and cannot be disabled.
  Users can no longer opt out of using the franz-go client in favor of the legacy Sarama client.
  The Sarama client and the feature gate will be removed completely after v0.143.0.

• extension/docker_observer: Upgrading Docker API version default from 1.24 to 1.44 (#​44279)

• pkg/ottl: Type of field profile.duration changes from time.Time to int64. (#​44397)

• receiver/azureeventhub: Promote Feature Gate receiver.azureeventhubreceiver.UseAzeventhubs to Beta (#​44335)

• receiver/k8slog: Update k8slogreceiver code-owners status and mark as unmaintained (#​44078)

• receiver/kafka: Remove deprecated topic and encoding (#​44568)

• receiver/kafka: receiver.kafkareceiver.UseFranzGo feature gate moved to Stable and is now always enabled (#​44564)
  The franz-go client is now the default and only Kafka client library for the Kafka receiver.
  The feature gate receiver.kafkareceiver.UseFranzGo has been promoted to Stable status and cannot be disabled.
  Users can no longer opt out of using the franz-go client in favor of the legacy Sarama client.
  The Sarama code and the feature gate will be removed completely after v0.143.0.

🚩 Deprecations 🚩

• receiver/prometheus: Add feature gate for extra scrape metrics in Prometheus receiver (#​44181)
  deprecation of extra scrape metrics in Prometheus receiver will be removed eventually.

🚀 New components 🚀

• connector/metricsaslogs: Add connector to convert metrics to logs (#​40938)
• extension/azure_encoding: [extension/azure_encoding] Introduce new component (#​41725)
  This change includes only overall structure, readme and configuration for a new component
• receiver/awslambda: Implementation of the AWS Lambda Receiver. (#​43504)
• receiver/macosunifiedlogging: Add a new receiver for macOS Unified Logging. (#​44089)
• receiver/macosunifiedlogging: Implementation of the macOS Unified Logging Receiver. (#​44089)

💡 Enhancements 💡

• connector/count: Support all attribute types in the count connector (#​43768)

• connector/routing: Avoid extra copy of all data during routing (#​44387)

• exporter/awss3: Support compression with ZSTD (#​44542)

• exporter/azuremonitor: Add additional mapping of standard OTel properties to builtin Application Insights properties (#​40598)

• exporter/cassandra: cassandraexporter: Upgrade cassandra library version (#​43691)
  Upgrade cassandra library version.

• exporter/elasticsearch: Updates the ecs mode span encode to include the span.kind attribute (#​44139)

• exporter/elasticsearch: add missing fields to struct so that they are populated in the respective elasticsearch index (#​44234)

• exporter/file: Add create_directory and directory_permissions options; exporter can automatically create parent directories (also honored by group_by) with configurable permissions. (#​44280)

  • New config: create_directory (bool) and directory_permissions (octal string, e.g. "0755").
  • When enabled, the exporter creates the parent directory of path on start.
  • group_by uses the configured permissions when creating per-attribute directories.

• exporter/googlecloudpubsub: Update to cloud.google.com/go/pubsub/v2. (#​44465)

• exporter/googlecloudpubsub: Add encoding extension support (#​42270, #​41834)
  Add encoding extension support for the payload on Pub/Sub. As having custom extensions means the Pub/Sub attributes
  cannot be auto discovered additional functionality has been added to set the message attributes.

• exporter/prometheus: Add without_scope_info to omit otel scope info from prometheus exporter metrics (#​43613)

• exporter/prometheus: Add support to exponential histograms (#​33703)

• exporter/signalfx: Makes sending tags from SignalFx Exporter configurable (#​43799)
  New optional configuration flag drop_tags has been added to SignalFx Exporter to allow users to disable tag metadata sending.
  This feature has been introduced due to a common issue among Splunk Observability customers when they're receiving more tags
  than allowed limit.

• extension/awslogs_encoding: Add more fields to AWS NLB logs at ELB extension (#​43757)

• extension/googlecloudlogentry_encoding: Add support for Proxy Network Load Balancer logs (#​44099)
  Add support for Proxy Network Load Balancer logs to the Google Cloud log entry encoding extension.
  This includes adding support for the new gcp.load_balancing.proxy_nlb attributes.
  See the README for more details.

• extension/headers_setter: Add support for chaining with other auth extensions via additional_auth configuration parameter. This allows combining multiple authentication methods, such as OAuth2 for bearer token authentication and custom headers for additional metadata. (#​43797)
  The additional_auth parameter enables the headers_setter extension to work in conjunction
  with other authentication extensions like oauth2client. The additional auth extension is called
  first to apply its authentication, then headers_setter adds its configured headers on top.

• extension/observer: Add container name, container ID, and container image to port endpoint (#​41309)

• pkg/ottl: Add ParseSeverity function to define mappings for log severity levels. (#​35778)

• pkg/ottl: Introduce CommunityID function to generate network hash (20-byte SHA1 digest) flow from the given source and destination {IP, port}, optionally protocol and seed values. (#​34062)

• pkg/ottl: Expand usage of literal into typed getters and avoid unnecessary work (#​44201)

• pkg/ottl: SliceToMap: add support to convert slices with non-map elements to maps (#​43099)

• processor/cumulativetodelta: Add support for exponential histograms (#​44106)

• processor/resourcedetection: Use osProfile.computerName for setting host.name in Azure resource detection processor (#​43959)

• processor/resourcedetectionprocessor/oraclecloud: Oracle Cloud (OCI) resource detection processor now differentiates between running off-platform (expected not to work), and encountering an error when running on-platform (expected to work) (#​42794)

  • Collectors not running on Oracle Cloud return an empty resource and no error, since we don't expect this case to work.
  • If the Oracle Cloud platform is detected but metadata fetch fails, the error is now logged and signaled to the processor, since we do expect this case to work.

• processor/tail_sampling: Add bytes_limiting policy type, sample based on the rate of bytes per second using a token bucket algorithm. (#​42509)

• processor/tail_sampling: Adaptive expected_new_traces_per_sec to improve performance lower bound (#​43561)

• receiver/googlecloudpubsub: Update to cloud.google.com/go/pubsub/v2. (#​44466)

• receiver/googlecloudpubsub: Adjusts the subscription regex to accommodate new project naming used for Google Trusted Partner Clouds. (#​43988)

• receiver/googlecloudpubsubpush: Add telemetry metrics to the component. (#​44422)

• receiver/googlecloudpubsubpush: Add implementation to googlecloudpubsubpush receiver. (#​44101)

• receiver/k8s_events: Allow more event types like Error and Critical which are typically used by applications when creating events. (#​43401)
  k8seventsreceiver allows event types Error and Critical in addition to the current Normal and Warning event types.

• receiver/kafka: Add support for exclude topics when consuming topics with a regex pattern (#​43782)

• receiver/prometheus: Support JWT Profile for Authorization Grant (RFC 7523 3.1) (#​44381)

• receiver/redis: Add support for redis.mode and redis.sentinel.* metrics (#​42365)

• receiver/systemd: Promote systemd receiver to alpha (#​33532)

• receiver/systemd: Scrape unit CPU time (#​44646)

🧰 Bug fixes 🧰

• cmd/opampsupervisor: Fix supervisor passthrough logs overflow by using bufio.Reader instead of bufio.Scanner (#​44127)

• cmd/opampsupervisor: Fix data race in remoteConfig field by using atomic pointer for thread-safe concurrent access (#​44173)

• connector/routing: Fix routing to default route when error occurs (#​44386)
  Before we used to send everything (even records match without error) to the default pipeline, |
  after this change only entries that return error will be "ignored" and if no other rule in the |
  table picks them will be sent to the default rule.

• exporter/clickhouse: Fix TLS configuration being ignored when only ca_file is provided and no cert/key files are set. (#​43911)
  This change ensures server-side TLS validation works correctly even without client certificates.

• exporter/elasticsearch: Fix CloudID parsing to correctly handle Elastic Cloud IDs when sent with multiple dollar sign separators (#​44306)
  The CloudID decoder was incorrectly using strings.Cut() which only splits on the first delimiter,
  causing malformed URLs when the decoded CloudID contained multiple $ separators. Changed to use
  strings.Split() to match the reference implementation from go-elasticsearch library.

• extension/awslogs_encoding: address the SIGSEGV occurring when processing control_message messages. (#​44231)

• extension/awslogs_encoding: Fix ALB log request_line parsing for valid formats and avoid errors (#​44233)

• pkg/ottl: Fixed OTTL grammar to treat the string literal "nil" as ordinary text instead of a nil value. (#​44374)

• pkg/ottl: Return errors when OTTL context setters receive values of the wrong type (#​40198)
  Introduces ctxutil.ExpectType and updates log, metric, and scope setters to surface type assertion failures.

• pkg/ottl: Fix TrimPrefix/TrimSuffix function name. (#​44630)
  This change also adds a featuregate "ottl.PanicDuplicateName" to control the behavior of panicing when duplicate
  names are registered for the same function.

• processor/k8sattributes: k8sattributesprocessor now respects semantic convention resolution order for service.namespace (#​43919)
  Previously, when service.namespace was included in the extract metadata configuration, the processor
  would incorrectly allow k8s.namespace.name to override explicitly configured service namespace values
  from OpenTelemetry annotations (e.g., resource.opentelemetry.io/service.namespace). Now the processor
  correctly follows the semantic convention resolution order, where annotation values take precedence over
  inferred Kubernetes namespace names.

• processor/k8sattributes: Fix incorrect pod metadata assignment when host.name contains a non-IP hostname (#​43938)
  The processor now correctly validates that host.name contains an IP address before using it for pod association.
  Previously, textual hostnames were incorrectly used for pod lookups, causing spans and metrics from one workload
  to receive metadata from unrelated pods that shared the same hostname.

• receiver/awsxray: Fix incorrect span kind when translating X-Ray segment to trace span with parent ID (#​44404)

• receiver/azuremonitor: Collect only supported aggregations for each metric (501 not implemented issue) (#​43648)
  Some metrics were not collected because we requested all available aggregation types. This led to 501 errors, as the Azure API returned responses indicating that certain aggregations were not implemented.
  We now use the supported aggregations field from each metric definition to filter and request only the aggregations that are actually supported.
  The user can expect less 501 errors in the logs and more metrics in the results.

• receiver/datadog: Utilizes thread safe LRU packages (#​42644)

• receiver/github: Adds corrections to span times when GitHub sends incorrect start and end times. (#​43180)

• receiver/libhoney: Allow single events and uncompressed requests (#​44026, #​44010)
  The receiver required events to be wrapped in an array before. The single-event format
  was allowed by Honeycomb's API so we have added it here.
  This fix also allows uncompressed requests again

• receiver/sqlquery: Fix a bug in the sqlqueryreceiver where an error is returned if the query returned a null value. This is now logged as a warning and logs with null values are ignored. (#​43984)

• receiver/systemd: This allows systemd receiver to be used in collector config (#​44420)

v0.140.1

Compare Source

v0.140.0

Compare Source

🛑 Breaking changes 🛑

• all: Latest supported k8s version is moved from 1.17 to 1.21. (#​43891)
• cmd/otelcontribcol: Removing unmaintained component extension/ecstaskobserver (#​43818)
• exporter/elasticsearch: Upgrade profiles proto to 1.9.0 (#​44031)
• extension/googlecloudlogentry_encoding: Parse cloud armor logs into log record attributes instead of placing it in the body as is. (#​43389)
• pkg/ottl: Upgrade profiles proto to 1.9.0 (#​44031)
• processor/filter: Upgrade profiles proto to 1.9.0 (#​44031)
• processor/tail_sampling: Replace policy latency metric with total time spent executing specific sampling policy. (#​42620)
  The existing latency metric was misleading and expensive to compute. The new cpu time metric can be used to find expensive policies instead.
• receiver/pprof: Upgrade profiles proto to 1.9.0 (#​44031)
• receiver/prometheus: The prometheus receiver no longer adjusts the start time of metrics by default. (#​43656)
  Disable the receiver.prometheusreceiver.RemoveStartTimeAdjustment | feature gate to temporarily re-enable this functionality. Users that need | this functionality should migrate to the metricstarttime processor, | and use the true_reset strategy for equivalent behavior.

🚀 New components 🚀

• pkg/translator/pprof: move code from receiver/pprofreceiver to pkg/translator/pprof (#​43975)
  pprof is an established format. For a better integration with OTel move code to a dedicated component.
• receiver/awslambda: Add scaffolding for the new awslambda receiver, for event-triggered receiving of data from S3 and CloudWatch. (#​43504)
• receiver/googlecloudpubsubpush: Add skeleton to the google pub sub push receiver. (#​43503)
• receiver/systemd: Report active state of systemd units. (#​33532)
• receiver/yanggrpc: New component YANG gRPC (#​43840)

💡 Enhancements 💡

• exporter/azureblob: Added serial_num_enabled and time_parser_enabled options to blob_name_format in Azure Blob Exporter to control random serial number appending and time parsing behavior. (#​43603)

• exporter/elasticsearch: Add support for latest OTEL SemConv version and fix Elasticsearch exporter ECS mapping for message.destination.name which is different for Elastic spans or transactions (#​43805, #​43806)

• exporter/elasticsearch: Add helpful error hint for illegal_argument_exception when using OTel mapping mode with Elasticsearch < 8.12 (#​39282)
  When using OTel mapping mode (default from v0.122.0) with Elasticsearch versions < 8.12,
  the exporter now provides a more descriptive error message explaining that OTel mapping mode
  requires Elasticsearch 8.12+ and suggests either upgrading Elasticsearch or using a different
  mapping mode. This helps users who encounter the generic illegal_argument_exception error
  understand the root cause and resolution steps.

• exporter/googlecloudstorage: Add googlecloudstorageexporter to the contrib distribution (#​44063)

• exporter/kafka: Adds a new configuration option to the Kafka exporter to control the linger time for the producer. (#​44075)
  Since franz-go now defaults to 10ms, it's best to allow users to configure this option to suit their needs.

• extension/datadog: Adds collector resource attributes to collector metadata payload (#​43979)
  The Collector's resource attributes can be set under service::telemetry::resource.

• extension/encoding: Add most of the AWS ELB fields to the AWSLogsEncoding. (#​43757)

• receiver/datadog: Adding log telemetry functionality to the existing datadog receiver component. (#​43841)

• receiver/github: Add include_span_events for GitHub Workflow Runs and Jobs for enhanced troubleshooting (#​43180)

• receiver/journald: Add root_path and journalctl_path config for running journald in a chroot (#​43731)

• receiver/prometheusremotewrite: Skip emitting empty metrics. (#​44149)

• receiver/prometheusremotewrite: prometheusremotewrite receiver now accepts metric type unspcified histograms. (#​41840)

• receiver/redis: Add redis metrics that are present in telegraf: cluster_enabled, tracking_total_keys, used_memory_overhead, used_memory_startup (#​39859)

• receiver/splunkenterprise: added pagination for search cases which may return more than the default 100 results (#​43608)

• receiver/webhookevent: Allow configuring larger webhook body size (#​43544)
  The receiver allows configuration a larger body buffer if needed.
  It also returns an error if the body exceeds the configured limit.

🧰 Bug fixes 🧰

• cmd/opampsupervisor: Redacts HTTP headers in debug message (#​43781)

• connector/datadog: Datadog connector no longer stalls after a downstream component errors (#​43980)

• exporter/awsxray: Fix conversion of the inProgress attribute into a Segment field instead of metadata (#​44001)

• exporter/datadog: Fix a panic from a race condition between exporter shutdown and trace export (#​44068)

• exporter/elasticsearch: Handle empty histogram buckets to not result in an invalid datapoint error. (#​44022)

• exporter/elasticsearch: Update the ecs mode span encode to correctly encode span.links ids as trace.id and span.id (#​44186)

• exporter/elasticsearch: Improve error message when an invalid Number data point is received. (#​39063)

• exporter/loadbalancing: Ensure loadbalancing child exporters use the OTLP type so backend creation succeeds (#​43950)

• exporter/stef: Fix STEF connection creation bug (#​44048)
  On some rare occasions due to a bug STEF exporter was incorrectly disconnecting just | created STEF connection causing connection error messages in the log. This fixes the bug.

• extension/bearertokenauth: Remove error messages `fsnoti