open-telemetry/opentelemetry-collector-contrib v0.109.0

on GitHub

The OpenTelemetry Collector Contrib contains everything in the opentelemetry-collector release, be sure to check the release notes there as well.

End User Changelog

🛑 Breaking changes 🛑

• clickhouseexporter: Upgrade trace SpanKind and StatusCode string values (#34799)
  This change updates the output of the trace SpanKind and StatusCode fields to be consistent
  with the specification's enum values. While this change will not break any deployments, it may affect
  queries dependent on the old enum names.

  For more details on old->new values, see this related PR:
  open-telemetry/opentelemetry-collector#6250

• spanmetricsconnector: Improve consistency between metrics generated by spanmetricsconnector. Added traces.span.metrics as default namespace (#33227, #32818)
  Default namespace for the generated metrics is traces.span.metrics now. | The deprecated metrics are: calls, duration and events. | The feature flag connector.spanmetrics.legacyLatencyMetricNames was added to revert the behavior.

• servicegraphconnector: Fix histogram metrics miss unit (#34511)
  All metrics will remove the suffix _seconds. It will not introduce breaking change if users use | prometheusexporter or prometheusremotewriteexporter to exporter metrics in pipeline. | In some cases, like using clickhouseexporter(save data in native OTLP format), it will be a breaking change. | Users can use transformprocessor to add back this suffix.

• gitproviderreceiver: The Git Provider Receiver has been renamed to GitHub Receiver. (#34731)
  This rename was to better match the OpenTelemetry semantic convention changes in v1.27.0
  and allows for Traces and Log signals to be added to this component. Traces
  as mentioned in issue #27460 will be added in a future release to this component.

• ottl: Remove tracing from OTTL due to performance concerns (#34910)

• vcenterreceiver: Updated units on several metrics to be more in line with documented semantics. (#34946)
  The following are the metrics with their respective unit changes:

  • vcenter.datacenter.cpu.limit ({MHz} -> MHz)
  • vcenter.cluster.cpu.limit ({MHz} -> MHz)
  • vcenter.cluster.cpu.effective ({MHz} -> MHz)
  • vcenter.cluster.vsan.operations ({operations/sec} -> {operations/s})
  • vcenter.cluster.vsan.congestions ({congestions/sec} -> {congestions/s})
  • vcenter.host.network.packet.error.rate ({errors/sec} -> {errors/s})
  • vcenter.host.network.packet.rate ({packets/sec} -> {packets/s})
  • vcenter.host.network.packet.drop.rate ({packets/sec} -> {packets/s})
  • vcenter.host.vsan.operations ({operations/sec} -> {operations/s})
  • vcenter.host.vsan.congestions ({congestions/sec} -> {congestions/s})
  • vcenter.resource_pool.cpu.usage ({MHz} -> MHz)
  • vcenter.vm.network.throughput (By/sec -> By/s)
  • vcenter.vm.network.packet.rate ({packets/sec} -> {packets/s})
  • vcenter.vm.vsan.operations ({operations/sec} -> {operations/s})

🚩 Deprecations 🚩

• instanaexporter: Marking instanaexporter module as deprecated (#34994)

💡 Enhancements 💡

• pkg/ottl: Added Decode() converter function (#32493)

• testbed: Add test case scenarios to handle large files to existing testbed. (#34288)

• filestorage: Add directory validation for compaction on-rebound (#35114)

• windowseventlogreceiver: Avoid rendering the whole event to obtain the provider name (#34755)

• datadogreceiver: add support for sketch metrics in Datadog receiver (#18278)

• datadogconnector: Optimize Datadog connector when there are many peer tags and connector.datadogconnector.NativeIngest is enabled (#34945)
  connector.datadogconnector.NativeIngest is currently enabled by default

• splunkhecexporter: Drop empty log events (#34871)
  Log records with no body are dropped by Splunk on reception
  as they contain no log message, albeit they may have attributes.

  This PR removes those logs from consideration to be exported.

  This is in tune with the behavior of splunkhecreceiver, which refuses HEC events with no event (#19769)

• elasticsearchexporter: Add exponential histogram support (#34813)

• elasticsearchexporter: Add span event support to traces OTel mapping mode (#34831)
  Span events are now supported in OTel mapping mode. They will be routed to logs-${data_stream.dataset}-${data_stream.namespace} if traces_dynamic_index::enabled is true.

• transformprocessor: Support aggregating metrics based on their attribute values and substituting the values with a new value. (#16224)

• kafkareceiver: Adds tunable fetch sizes to Kafka Receiver (#22741, #34431)
  Adds the ability to tune the minumum, default and maximum fetch sizes for the Kafka Receiver

• solarwindsapmsettingsextension: Added logic for refresh function (#27668)

• githubreceiver: Promote GitHub receiver metrics to alpha status. (#34960)

• googlecloudmonitoringreceiver: Enhancing the Google Cloud monitoring receiver to establish a client connection, scrape GCP Cloud Metrics, and transform them into an OpenTelemetry compatible format for pipeline processing. (#33762)

  • Implements client connection to Google Cloud Monitoring API.
  • Scrapes timeseries data based on configured metrics.
  • Converts the data into OpenTelemetry format for use in the pipeline.

• processor/interval: Support for gauge and summary metrics. (#34803)
  Only the last value of a gauge or summary metric is reported in the interval processor, instead of all values.

• kafkareceiver: Add support for encoding extensions in the Kafka receiver. (#33888)
  This change adds support for encoding extensions in the Kafka receiver. Loading extensions takes precedence over the internally supported encodings.

• datadogexporter: Add support for setting a custom log source from resource attribute datadog.log.source (#35051)

• opampextension: Adds the ability to configure the polling interval for the HTTP client. (#34749)

• otelarrowexporter: Add gRPC timeout propagation. (#34733)

• pkg/ottl: Add Sort function to sort array to ascending order or descending order (#34200)

• exporter/prometheusexpoter: Support for Prometheus Created Timestamps. (#32521)

• redactionprocessor: Add support for logs and metrics (#34479)

• exceptionsconnector,servicegraphconnector,spanmetricsconnector: Extract the getDimensionValue function as a common function. (#34627)

• sqlqueryreceiver: Support populating log attributes from sql query (#24459)

• opampsupervisor: Add new config parameter agent.health_check_port to allow configuring the port used by the agent healthcheck extension. (#34643)

🧰 Bug fixes 🧰

• apachereceiver: Fix panic on invalid endpoint configuration (#34992)

• deltatocumulative: drop bad samples (#34979)
  removes bad (rejected) samples from output. previously identified and metric-tracked those as such, but didn't actually drop them.

• elasticsearchexporter: Fix dynamic mapping for double values storing integers (#34680)

• elasticsearchexporter: Stringify attribute array values in metrics OTel mode (#35004)
  Elasticsearch TSDB does not support array dimensions. Workaround it by stringifying attribute array values in OTel mapping mode for metrics.

• fileconsumer: Fix bug where max_concurrent_files could not be set to 1. (#35080)

• servicegraphconnector: Fix incorrectly reversed latency settings for the server and client (#34933)

• hostmetricsreceiver: In filesystem scraper, do not prefix partitions when using the environment variable HOST_PROC_MOUNTINFO (#35043)

• splunkhecreceiver: Fix memory leak when the receiver is used for both metrics and logs at the same time (#34886)

• geoipprocessor: Avoid using internal empty attribute.Set pointer (#34882)

• pkg/stanza: Synchronize shutdown in stanza adapter (#31074)
  Stanza-based receivers should now flush all data before shutting down

• sqlserverreceiver: Fix bug where metrics were being emitted with the wrong database name resource attribute (#35036)

• signalfxexporter: Fix memory leak by re-organizing the exporter's functionality lifecycle (#32781)

• otlpjsonconnector: Handle OTLPJSON unmarshal error (#34782)

• mysqlreceiver: mysql client raise error when the TABLE_ROWS column is NULL, convert NULL to int64 (#34195)

• pkg/stanza: An operator configured with silent errors shouldn't log errors while processing log entries. (#35008)

• datadogexporter: Use correct hostname for logs when using Datadog Agent logs pipeline with a gateway deployment. (#35058)

API Changelog

🛑 Breaking changes 🛑

• pkg/ottl: Change the OTTL grammar to support expressing statements context via path names (#29017)
  The ottl.Path interface requires a new method: Context() string
• prometheusreceiver: Move the TargetAllocator configuration struct to an internal directory (#33146)

💡 Enhancements 💡

• pkg/ottl: Add support for localized time parsing into the coreinternal/timeutils (#32977)
• googlecloudmonitoringreceiver: Enhancing the Google Cloud monitoring receiver to establish a client connection, scrape GCP Cloud Metrics, and transform them into an OpenTelemetry compatible format for pipeline processing. (#33762)
  • Implements client connection to Google Cloud Monitoring API.
  • Scrapes timeseries data based on configured metrics.
  • Converts the data into OpenTelemetry format for use in the pipeline.