signalfx/splunk-otel-collector v0.140.0

on GitHub

This Splunk OpenTelemetry Collector release includes changes from the opentelemetry-collector v0.140.0
and the opentelemetry-collector-contrib v0.140.1 releases where appropriate.

🛑 Breaking changes 🛑

• (Splunk) ecs_task_observer: Removed discontinued ecs_task_observer extension. (#6936)
  The ecs_task_observer extension has been discontinued and is no longer available in the OpenTelemetry Collector Contrib repository
  and was removed via PR 43818.
• (Splunk) gateway: Remove SignalFx receiver from the default gateway config in favor of the HTTP forwarder extension (#6887)
  This change means that metrics and logs sent to the gateway in the SignalFx format
  will no longer be subject to the batch and memory limiter processors configured in
  the gateway pipelines. Please ensure any required memory limiting or batching is
  configured on the sending side (e.g. the agent). Please note that the signalfx receiver
  will eventually be deprecated and removed.
• (Contrib) receiver/ciscoos: Rename receiver component name from ciscoosreceiver to ciscoos to follow naming conventions. (#42647)
  Users must update their collector configuration from ciscoosreceiver/device to ciscoos/device.
• (Contrib) all: Latest supported k8s version is moved from 1.17 to 1.21. (#43891)
• (Contrib) extension/googlecloudlogentry_encoding: Parse cloud armor logs into log record attributes instead of placing it in the body as is. (#43389)
• (Contrib) pkg/ottl: Upgrade profiles proto to 1.9.0 (#44031)
• (Contrib) processor/filter: Upgrade profiles proto to 1.9.0 (#44031)
• (Contrib) 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.
• (Contrib) receiver/prometheus: The prometheus receiver no longer adjusts the start time of metrics by default. (#43656)

🚩 Deprecations 🚩

• (Splunk) scriptedinputsreceiver: Deprecate the scripted inputs receiver. (#6893)
• (Splunk) receiver/lightprometheus: The lightprometheus receiver has been deprecated and will be removed in a future release. (#6949)
  Please use the Prometheus or
  Simple Prometheus receiver instead.

🚀 New components 🚀

• (Splunk) exporter/prometheusremotewrite: Add the Prometheus Remote Write exporter to the distribution. (#6912)
  See README
  for more information.

💡 Enhancements 💡

• (Splunk) javaagent: Update Splunk Distribution of OpenTelemetry Java to version 2.21.1 (#6944)
• (Splunk) discovery: Add a new discovery bundle for Weaviate vector search engine (#6884)
  The discovery rule targets Weaviate instances exposing Prometheus metrics on the default port 2112.
• (Core) connector/forward: Add support for Profiles to Profiles (#14092)
• (Core) exporter/debug: Disable sending queue by default (#14138)
  The recently added sending queue configuration in Debug exporter was enabled by default and had a problematic default size of 1.
  This change disables the sending queue by default.
  Users can enable and configure the sending queue if needed.
• (Core) pkg/otelcol: This feature has been improved and tested; secure-by-default redacts configopaque values (#12369)
• (Contrib) 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.
• (Contrib) receiver/journald: Add root_path and journalctl_path config for running journald in a chroot (#43731)
• (Contrib) receiver/prometheusremotewrite: Skip emitting empty metrics. (#44149)
• (Contrib) receiver/prometheusremotewrite: Receiver now accepts metric type unspecified histograms. (#41840)
• (Contrib) receiver/redis: Add redis metrics that are present in telegraf: cluster_enabled, tracking_total_keys, used_memory_overhead, used_memory_startup (#39859)
• (Contrib) receiver/splunkenterprise: Added pagination for search cases which may return more than the default 100 results (#43608)

🧰 Bug fixes 🧰

• (Splunk) smartagent/docker-container-stats: Use client API version negotiation instead of fixed version (#6941)
  The smartagent/docker-container-stats monitor was using a fixed API version of v1.24,
  which is not compatible with version v29 of the Docker engine. This change updates the
  monitor to use API version negotiation, allowing it to work with a wider range of Docker
  engine versions.
• (Core) all: Ensure service.instance.id is the same for all the signals when it is autogenerated. (#14140)
• (Contrib) exporter/loadbalancing: Ensure loadbalancing child exporters use the OTLP type so backend creation succeeds (#43950)
• (Contrib) extension/bearertokenauth: Remove error messages fsnotify: can't remove non-existent watch when watching kubernetes SA tokens. (#44104)
• (Contrib) processor/k8sattributes: The fix is on k8sattributes processor to only set k8s.pod.ip attribute when it is requested in the extract.metadata configuration. (#43862)
  Previously, the k8s.pod.ip attribute was always populated, even if it was not included in the extract.metadata list.
  This fix ensures that k8s.pod.ip is set only when explicitly requested, aligning the processor behavior with configuration expectations.
• (Contrib) receiver/sqlserver: Resolved inaccurate data sampling in query metrics collection. (#44303)
• (Contrib) receiver/sqlserver: Fix incorrect logic in query metrics window calculation. (#44162)
• (Contrib) receiver/sqlserver: Fixed a bug in effective value calculation of lookback time in top query collection. (#43943)

Docker Image Manifests:

• Linux (amd64, arm64, ppc64le) and Windows (2019 amd64, 2022 amd64):
  • quay.io/signalfx/splunk-otel-collector:0.140.0
  • digest: sha256:147b26e5dbf7358557089a50742c17e6581e2ac481f3f38506097b9a65f11eab