open-telemetry/opentelemetry-collector-contrib v0.83.0

on GitHub

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

User Facing Changes

🛑 Breaking changes 🛑

• receiver/k8scluster: Unify predefined and custom node metrics. (#24776)

  • Update metrics description and units to be consistent
  • Remove predefined metrics definitions from metadata.yaml because they are controlled by node_conditions_to_report
    and allocatable_types_to_report config options.

• prometheusexporter: Remove invalid unit translations from the prometheus exporters (#24647)

• receiver/prometheusexec: Removes the deprecated prometheus_exec receiver (#24740)

🚀 New components 🚀

• datadogconnector: This is a new component that computes Datadog APM Stats in the event that trace pipelines are sampled. (#19740)
  This component replaces the Datadog processor

• gitproviderreceiver: Add the skeleton for the new gitproviderreceiver in development with accompanying github scraper. (#22028)

💡 Enhancements 💡

• azuredataexplorerexporter: Add support for managed identity. This enables users to not use Key based authentication (#21924)

• awsemfexporter: Add awsemf.nodimrollupdefault feature gate to aws emf exporter (#23997)
  Enabling the awsemf.nodimrollupdefault will cause the AWS EMF Exporter to use the NoDimensionRollup configuration
  setting by default instead of ZeroAndSingleDimensionRollup.

• awss3exporter: add Sumo Logic Installed Collector marshaler (#23212)

• receiver/collectdreceiver: Migrate from opencensus to pdata, change collectd, test to match pdata format. (#20760)

• datasetexporter: Make duration of shutdown procedure configurable to minimise data losses. (#24415)

• datasetexporter: Make sure serverHost field is correctly and always populated on the DataSet events. For more information and available configuration options, please refer to the plugin readme file. (#20660, #24415)

• datadogreceiver: add datadog trace and span id (#23057)

• pkg/ottl: Add support for using addition and subtraction with time and duration (#22009)

• transformprocessor: Add extract_count_metric OTTL function to transform processor (#22853)

• transformprocessor: Add extract_sum_metric OTTL function to transform processor (#22853)

• prometheusreceiver: Don't drop histograms without buckets (#22070)

• pkg/ottl: Add a new Function Getter to the OTTL package, to allow passing Converters as literal parameters. (#22961)
  Currently OTTL provides no way to use any defined Converter within another Editor/Converter.
  Although Converters can be passed as a parameter, they are always executed and the result is what is actually passed as the parameter.
  This allows OTTL to pass Converters themselves as a parameter so they can be executed within the function.

• resourcedetectionprocessor: GCP resource detection processor can automatically add gcp.gce.instance.hostname and gcp.gce.instance.name attributes. (#24598)

• splunkhecexporter: Add heartbeat check while startup and new config param, heartbeat/startup (defaults to false). This is different than the healtcheck_startup, as the latter doesn't take token or index into account. (#24411)

• hostmetricsreceiver: Report logical and physical number of CPUs as metric. (#22099)
  Use the system.cpu.logical.count::enabled and system.cpu.physical.count::enabled flags to enable them

• k8sclusterreceiver: Allows disabling metrics and resource attributes (#24568)

• k8sclusterreceiver: Reduce memory utilization (#24769)

• k8sattributes: Added k8s.cluster.uid to k8sattributes processor to add cluster uid (#21974)

• resourcedetectionprocessor: Collect heroku metadata available instead of exiting early. Log at debug level if metadata is missing to help troubleshooting. (#25059)

• hostmetricsreceiver: Improved description of the system.cpu.utilization metrics. (#25115)

• cmd/mdatagen: Avoid reusing the same ResourceBuilder instance for multiple ResourceMetrics (#24762)

• resourcedetectionprocessor: Add detection of os.description to system detector (#24541)

• filelogreceiver: Bump 'filelog.allowHeaderMetadataParsing' feature gate to beta (#18198)

• receiver/purefareceiver: implement the custom label fa_array_name to act as a pretty label for metrics received. (#23889, #21248, #22027)

• receiver/prometheusreceiver: Add config report-extra-scrape-metrics to report additional prometheus scraping metrics (#21040)
  Emits additional metrics - scrape_body_size_bytes, scrape_sample_limit, scrape_timeout_seconds. scrape_body_size_bytes metric can be used for checking failed scrapes due to body-size-limit.

• receiver/sqlquery: Set ObservedTimestamp on collected logs (#23776)

• exporter/awss3exporter: Allow custom endpoints to be configured for exporting spans (#21833)

• cmd/telemetrygen: Add ability to set custom path to endpoint. (#24551)

• telemetrygen: Adds batch option to configure whether to batch traces, and size option to configure minimum size in MB of each trace for load testing. (#9597)

• webhookreceiver: Add an optional config setting to set a required header that all incoming requests must provide (#24270)

• extension/jaegerremotesampling: gRPC remote source usage in jaegerremotesampling extension propagates HTTP headers if set in gRPC client config (#24414)

• extension/jaegerremotesampling: gRPC remote source usage in jaegerremotesampling extension supports optional caching via existing reload_interval config (#24840)

🧰 Bug fixes 🧰

• receiver/sshcheck: Add the SSH endpoint as a resource attribute (#24441)

• awsemfexporter: Enforce time to live on metric data that is stored for the purpose of cumulative to delta conversions within EMF Exporter (#25058)
  This change fixes a bug where the cache used to store metric information for cumulative to delta
  conversions was not enforcing its time to live. This could cause excessive memory growth in certain scenarios which could
  lead to OOM failures for Collector. To properly fix this issue package global metric caches were removed and replaced
  with caches that are unique per emf exporter. A byproduct of this change is that no two emf exporters within an
  Collector will share a caches leading to more accurate cumulative to delta conversions.

• awsemfexporter: Add retain_initial_value_of_delta_metric to translateOTelToGroupedMetric, allowing the initial set of metrics to be published (#24051)

• carbonreceiver: Fix Carbon receiver obsrecv operations memory leak (#24275)
  The carbonreceiver has a memory leak where it will repeatedly open new obsrecv operations but not close them afterwards. Those operations eventually create a burden.

  The fix is to make sure the receiver only creates an operation per interaction over TCP.

• datadogexporter: Populate OTLP resource attributes in Datadog logs. Changes mapping for jvm.loaded_classes from process.runtime.jvm.classes.loaded to process.runtime.jvm.classes.current_loaded. (#24674)

• datadogexporter: Fix the population of Datadog system.* metrics. Ensure the average is within [min, max] in histograms. (#25071)
  The minimum and maximum estimation is only used when the minimum and maximum are not available in the OTLP payload or this is a cumulative payload.

• pkg/stanza: Create a new decoder for each TCP/UDP connection to prevent concurrent write to buffer. (#24980)

• exporter/kafkaexporter: Fixes a panic when SASL configuration is not present (#24797)

• lokitranslator, lokiexporter: Fixes a panic that occurred during the promotion of nested attributes containing dots to labels (#25125)

• awsxrayexporter: Fix X-Ray Segment status code and exception translations. (#24381)

• receiver/haproxy: Make sure emitted resource metrics have distinct resources by default (#24921)
  This is done by enabling and renaming the following resource attributes:

  • proxy_name -> haproxy.proxy_name
  • service_name -> haproxy.service_name

• receiver/k8sobjects: Fix bug where duplicate data would be ingested for watch mode if the client connection got reset. (#24806)

• datadogreceiver: Fixed NPE on failed to decode message path (#24562)

• zipkinreceiver: Respects zipkin's serialised status tags to be converted to span status (#14965)

• datadogexporter: Correctly set metrics exporter capabilities to state that it mutates data (#24908)
  This could lead to random panics if using more than one metrics exporter.

• processor/resourcedetection: Do not drop all system attributes if host.id cannot be fetched. (#24669)

• signalfxexporter: convert vmpage_io* translated metrics to pages (#25064)

• splunkhecreceiver: aligns success resp body w/ splunk enterprise (#19219)
  changes resp from plaintext "ok" to json {"text"："success", "code"：0}

Go API Changes

🛑 Breaking changes 🛑

• exporter/clickhouse: Change the type of Config.Password to be configopaque.String (#17273)
• all: Remove go 1.19 support, bump minimum to go 1.20 and add testing for 1.21 (#8207)
• solacereceiver: Move model package to the internal package (#24890)
• receiver/statsdreceiver: Move protocol and transport packages to internal (#24892)
• filterprocessor: Unexport Strict and Regexp (#24845)
• mdatagen: Rename the mdatagen sum field aggregation to aggregation_temporality (#16374)
• metricstransformprocessor: Unexport elements of the Go API of the processor (#24846)
• mezmoexporter: Unexport the MezmoLogLine and MezmoLogBody structs (#24842)
• pkg/stanza: Remove deprecated 'fileconsumer.FileAttributes' (#24688)
• pkg/stanza: Remove deprecated 'fileconsumer.EmitFunc' (#24688)
• pkg/stanza: Remove deprecated fileconsumer.Finder (#24688)
• pkg/stanza: Remove deprecated fileconsumer.BaseSortRule and fileconsumer.SortRuleImpl (#24688)
• pkg/stanza: Remove deprecated 'fileconsumer.Reader' (#24688)

🚩 Deprecations 🚩

• pkg/stanza: Deprecate helper.Encoding and helper.EncodingConfig.Build (#24980)
• pkg/stanza: Deprecate fileconsumer MatchingCriteria in favor of new matcher package (#24853)

💡 Enhancements 💡

• changelog: Generate separate changelogs for end users and package consumers (#24014)
• splunkhecexporter: Add heartbeat check while startup and new config param, heartbeat/startup (defaults to false). This is different than the healtcheck_startup, as the latter doesn't take token or index into account. (#24411)
• k8sclusterreceiver: Allows disabling metrics and resource attributes (#24568)
• cmd/mdatagen: Avoid reusing the same ResourceBuilder instance for multiple ResourceMetrics (#24762)

🧰 Bug fixes 🧰

• splunkhecreceiver: aligns success resp body w/ splunk enterprise (#19219)
  changes resp from plaintext "ok" to json {"text"："success", "code"：0}