open-telemetry/opentelemetry-collector-contrib v0.94.0

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 🛑

• servicegraphprocessor: removed deprecated component, use the servicegraph connector instead. (#26091)

• datadogconnector: Enable feature gate connector.datadogconnector.performance by default. (#30829)
  See https://github.com/open-telemetry/opentelemetry-collector-contrib/tree/main/connector/datadogconnector#feature-gate-for-performance for caveats of this feature gate.

• datadogprocessor: Delete datadogprocessor which has been deprecated since v0.84.0 (#31026)
  Use datadogconnector instead.

• kafkareceiver: standardizes the default topic name for metrics and logs receivers to the same topic name as the metrics and logs exporters of the kafkaexporter (#27292)
  If you are using the Kafka receiver in a logs and/or a metrics pipeline
  and you are not customizing the name of the topic to read from with the topic property,
  the receiver will now read from otlp_logs or otlp_metrics topic instead of otlp_spans topic.
  To maintain previous behavior, set the topic property to otlp_spans.

• pkg/stanza: Entries are no longer logged during error conditions. (#26670)
  This change is being made to ensure sensitive information contained in logs are never logged inadvertently.
  This change is a breaking change because it may change user expectations. However, it should require
  no action on the part of the user unless they are relying on logs from a few specific error cases.

• pkg/stanza: Invert recombine operator's 'overwrite_with' default value. (#30783)
  Previously, the default value was oldest, meaning that the recombine operator should emit the
  first entry from each batch (with the recombined field). However, the actual behavior was inverted.
  This fixes the bug but also inverts the default setting so as to effectively cancel out the bug fix
  for users who were not using this setting. For users who were explicitly setting overwrite_with,
  this corrects the intended behavior.

🚩 Deprecations 🚩

• skywalkingexporter: Mark the component as unmaintained. If we don't find new maintainers, it will be deprecated and removed. (#23796)

🚀 New components 🚀

• sumologicextension: add configuration and readme (#29601)
• failoverconnector: Refactor of connector to seperate concerns between managing indexes and core failover component (#20766)
• otelarrow: Skeleton of new OpenTelemetry Protocol with Apache Arrow Receiver (#26491)
• processor/interval: Adds the initial structure for a new processor that aggregates metrics and periodically forwards the latest values to the next component in the pipeline. (#29461)
  As per the CONTRIBUTING.md recommendations, this PR only creates the basic structure of the processor. The concrete implementation will come as a separate followup PR

💡 Enhancements 💡

• receiver/journald: add a new config option "all" that turns on full output from journalctl, including lines that are too long. (#30920)

• pkg/stanza: Add support in a header configuration for json array parser. (#30321)

• awss3exporter: Add the ability to export trace/log/metrics in OTLP ProtoBuf format. (#30682)

• dockerobserver: Upgrading Docker API version default from 1.22 to 1.24 (#30900)

• filterprocessor: move metrics from OpenCensus to OpenTelemetry (#30736)

• groupbyattrsprocessor: move metrics from OpenCensus to OpenTelemetry (#30763)

• datadogconnector: Add trace configs that mirror datadog exporter (#30787)
  ignore_resources: disable certain traces based on their resource name
  span_name_remappings: map of datadog span names and preferred name to map to
  span_name_as_resource_name: use OTLP span name as datadog operation name
  compute_stats_by_span_kind: enables an additional stats computation check based on span kind
  peer_tags_aggregation: enables aggregation of peer related tags
  trace_buffer: specifies the buffer size for datadog trace payloads

• elasticsearchexporter: Add mapping.mode: raw configuration option (#26647)
  Setting mapping.mode: raw in the Elasticsearch exporter's configuration
  will result logs and traces being indexed into Elasticsearch with their
  attribute fields directly at the root level of the document instead inside an
  Attributes object. Similarly, this setting will also result in traces being
  indexed into Elasticsearch with their event fields directly at the root level
  of the document instead of inside an Events object.

• loadbalancingexporter: Optimize metrics and traces export (#30141)

• gitproviderreceiver: Add pull request metrics (#22028)

  • git.repository.pull_request.open.count
  • git.repository.pull_request.open.time
  • git.repository.pull_request.merged.count
  • git.repository.pull_request.merged.time
  • git.repository.pull_request.approved.time

• all: Add component.UseLocalHostAsDefaultHost feature gate that changes default endpoints from 0.0.0.0 to localhost (#30702)
  This change affects the following components:

  • extension/awsproxy
  • extension/health_check
  • extension/jaegerremotesampling
  • internal/aws/proxy
  • processor/remotetap
  • receiver/awsfirehose
  • receiver/awsxray
  • receiver/influxdb
  • receiver/jaeger
  • receiver/loki
  • receiver/opencensus
  • receiver/sapm
  • receiver/signalfx
  • receiver/skywalking
  • receiver/splunk_hec
  • receiver/zipkin
  • receiver/zookeeper

• googlepubsubreceiver: Add support for GoogleCloud logging encoding (#29299)

• processor/resourcedetectionprocessor: Detect Azure cluster name from IMDS metadata (#26794)

• processor/transform: Add copy_metric function to allow duplicating a metric (#30846)

🧰 Bug fixes 🧰

• basicauthextension: Accept empty usernames. (#30470)
  Per https://datatracker.ietf.org/doc/html/rfc2617#section-2, username and password may be empty strings ("").
  The validation used to enforce that usernames cannot be empty.

• servicegraphconnector: update prefix to match the component type (#31023)

• datadog/connector: Create a separate connector in the Datadog connector for the trace-to-metrics and trace-to-trace pipelines. It should reduce the number of conversions we do and help with Datadog connector performance. (#30828)
  Simplify datadog/connector with two separate connectors in trace-to-metrics pipeline and trace-to-trace pipeline.

• datadogreceiver: Set AppVersion to allow Datadog version property to transform properly to service.version resource attribute (#30225)

• cmd/opampsupervisor: Fix memory leak on shutdown (#30438)

• exporter/datadog: Fixes a bug where empty histograms were not being sent to the backend in the distributions mode. (#31019)

• pkg/ottl: Fix parsing of string escapes in OTTL (#23238)

• pkg/stanza: Recombine operator should always recombine partial logs (#30797)
  Previously, certain circumstances could result in partial logs being emitted without any
  recombiniation. This could occur when using is_first_entry, if the first partial log from
  a source was emitted before a matching "start of log" indicator was found. This could also
  occur when the collector was shutting down.

• pkg/stanza: Fix bug where recombine operator's 'overwrite_with' condition was inverted. (#30783)

• exporter/signalfx: Use "unknown" value for the environment correlation calls as fallback. (#31052)
  This fixed the APM/IM correlation in the Splunk Observability UI for the users that send traces with no "deployment.environment" resource attribute value set.

• namedpipereceiver: Fix SIGSEGV when named pipe creation fails (#31088)

API Changes

🚩 Deprecations 🚩

• testbed: Deprecate testbed.GetAvailablePort in favor of testutil.GetAvailablePort (#30811)
  Move healthcheckextension to use testutil.GetAvailablePort

🚀 New components 🚀

• pkg_sampling: Package of code for parsing OpenTelemetry tracestate probability sampling fields. (#29738)