Grafana Alloy Updates & Release Notes

Features

• Update the prometheus.exporter.windows component to version v0.30.7. This adds new metrics to the dns collector. (@dehaansa)

Bugfixes

• Update the prometheus.exporter.windows component to version v0.30.7. This fixes an error with the exchange collector and terminal_services collector (@dehaansa)

• Fix loki.source.firehose to propagate specific cloudwatch event timestamps when useIncomingTs is set to true. (@michaelPotter)

• Fix elevated CPU usage when using some otelcol components due to debug logging. (@thampiotr)

Other changes

• Upgrade otelcol components from OpenTelemetry v0.125.0 to v0.126.0 (@dehaansa):
  • [pkg/ottl] Add support for HasPrefix and HasSuffix functions.
  • [pkg/configtls] Add trusted platform module (TPM) support to TLS authentication for all otelcol components supporting TLS.
  • [otelcol.connector.spanmetrics] Add calls_dimension and histogram:dimension blocks for configuring additional dimensions for traces.span.metrics.calls and traces.span.metrics.duration metrics.
  • [otelcol.exporter.datadog] Enable instrumentation_scope_metadata_as_tags by default.
  • [otelcol.exporter.kafka] support configuration of compression level in producer configuration.
  • [otelcol.processor.tailsampling] invert sample and inverted not sample decisions deprecated, use the drop policy instead to explicitly not sample traces. …

Breaking changes

• The prometheus.exporter.windows component has been update to version v0.30.6. This update includes a significant rework of the exporter and includes some breaking changes. (@dehaansa)

  • The msmq and service collectors can no longer be configured with a WMI where clause. Any filtering previously done in a where clause will need to be done in a prometheus.relabel component.
  • The service collector no longer provides enable_v2_collector and use_api configuration options.
  • The mscluster_* and netframework_* collectors are now replaced with one mscluster and netframework collector that allows you to enable the separate metric groupings individually.
  • The teradici_pcoip and vmware_blast collectors have been removed from the exporter.

• The prometheus.exporter.oracledb component now embeds the oracledb_exporter from oracle instead of the deprecated oracledb_exporter from iamseth for collecting metrics from an OracleDB server: (@wildum)

  • The arguments username, password, default_metrics, and custom_metrics are now supported.
  • The previously undocumented argument custom_metrics is now expecting a list of paths to custom metrics files.
  • The following metrics are no longer available by default: oracledb_sessions_activity, oracledb_tablespace_free_bytes …

Breaking changes

• Removed open_port and executable_name from top level configuration of Beyla component. Removed enabled argument from network block. (@marctc)

• Breaking changes from the OpenTelemetry Collector v0.122 update: (@wildum)

  • otelcol.exporter.splunkhec: min_size_items and max_size_items were replaced by min_size, max_size and sizer in the batcher block to allow users to configure the size of the batch in a more flexible way.
  • The telemetry level of Otel components is no longer configurable. The level argument in the debug_metrics block is kept to avoid breaking changes but it is not used anymore.
  • otelcol.processor.tailsampling changed the unit of the decision timer metric from microseconds to milliseconds. (change unit of otelcol_processor_tail_sampling_sampling_decision_timer_latency)
  • otelcol.processor.deltatocumulative: rename otelcol_deltatocumulative_datapoints_processed to otelcol_deltatocumulative_datapoints and remove the metrics otelcol_deltatocumulative_streams_evicted, otelcol_deltatocumulative_datapoints_dropped and otelcol_deltatocumulative_gaps_length.
  • The regex attribute was removed from otelcol.processor.k8sattributes. The extract-patterns function from otelcol.processor.transform can be used instead.
  • The default value of metrics_flush_interval in otelcol.connector.servicegraph was changed from 0s to 60s. …

Bugfixes

• Fixed an issue where the otelcol.exporter.awss3 could not be started with the sumo_ic marshaler. (@wildum)

• Update jfr-parser dependency to v0.9.3 to fix jfr parsing issues in pyroscope.java. (@korniltsev)

• Fixed an issue where passing targets from some standard library functions was failing with target::ConvertFrom error. (@thampiotr)

• Fixed an issue where indexing targets as maps (e.g. target["foo"]) or objects (e.g. target.foo) or using them with certain standard library functions was resulting in expected object or array, got capsule error under some circumstances. This could also lead to foreach evaluation failed errors when using the foreach configuration block. (@thampiotr)

• Update prometheus.write.queue to reduce memory fragmentation and increase sent throughput. (@mattdurham)

• Fixed an issue where the otelcol.exporter.kafka component would not start if the encoding was specific to a signal type. (@wildum)

Breaking changes

• (Experimental) In prometheus.write.queue changed parallelism from attribute to a block to allow for dynamic scaling. (@mattdurham)

• Remove tls_basic_auth_config_path attribute from prometheus.exporter.mongodb configuration as it does not configure TLS client behavior as previously documented.

• Remove encoding and encoding_file_ext from otelcol.exporter.awss3 component as it was not wired in to the otel component and Alloy does not currently integrate the upstream encoding extensions that this would utilize.

Features

• Add a otelcol.receiver.tcplog component to receive OpenTelemetry logs over a TCP connection. (@nosammai)

• (Public preview) Add otelcol.receiver.filelog component to read otel log entries from files (@dehaansa)

• (Public preview) Add a otelcol.processor.cumulativetodelta component to convert metrics from cumulative temporality to delta. (@madaraszg-tulip)

• (Experimental) Add a stage.windowsevent block in the loki.process component. This aims to replace the existing stage.eventlogmessage. (@wildum)

• Add pyroscope.relabel component to modify or filter profiles using Prometheus relabeling rules. (@marcsanmi)

• (Experimental) A new foreach block which starts an Alloy pipeline for each item inside a list. (@wildum, @thampiotr, @ptodev)

Enhancements

• Upgrade to OpenTelemetry Collector v0.119.0 (@dehaansa): …

Breaking changes

• Upgrade to OpenTelemetry Collector v0.116.0:

  • otelcol.processor.tailsampling: Change decision precedence when using and_sub_policy and invert_match. For more information, see the release notes for Alloy 1.6.

Features

• Add support for TLS to prometheus.write.queue. (@mattdurham)

• Add otelcol.receiver.syslog component to receive otel logs in syslog format (@dehaansa)

• Add support for metrics in otelcol.exporter.loadbalancing (@madaraszg-tulip)

• Add add_cloudwatch_timestamp to prometheus.exporter.cloudwatch metrics. (@captncraig)

• Add support to prometheus.operator.servicemonitors to allow endpointslice role. (@yoyosir)

• Add otelcol.exporter.splunkhec allowing to export otel data to Splunk HEC (@adlotsof)

• Add otelcol.receiver.solace component to receive traces from a Solace broker. (@wildum)

• Add otelcol.exporter.syslog component to export logs in syslog format (@dehaansa)

• (Experimental) Add a database_observability.mysql component to collect mysql performance data. (@cristiangreco & @matthewnolf)

• Add otelcol.receiver.influxdb to convert influx metric into OTEL. (@EHSchmitt4395)

• Add a new /-/healthy endpoint which returns HTTP 500 if one or more components are unhealthy. (@ptodev)…

Enhancements

• Logs from underlying clustering library memberlist are now surfaced with correct level (@thampiotr)

• Allow setting informer_sync_timeout in prometheus.operator.* components. (@captncraig)

• For sharding targets during clustering, loki.source.podlogs now only takes into account some labels. (@ptodev)

• Improve instrumentation of pyroscope.relabel component. (@marcsanmi)

Bugfixes

• Fixed an issue in the pyroscope.write component to prevent TLS connection churn to Pyroscope when the pyroscope.receive_http clients don't request keepalive (@madaraszg-tulip)

• Fixed an issue in the pyroscope.write component with multiple endpoints not working correctly for forwarding profiles from pyroscope.receive_http (@madaraszg-tulip)

• Fixed a few race conditions that could lead to a deadlock when using import statements, which could lead to a memory leak on /metrics endpoint of an Alloy instance. (@thampiotr)

• Fix a race condition where the ui service was dependent on starting after the remotecfg service, which is not guaranteed. (@dehaansa & @erikbaranowski)

• Fixed an issue in the otelcol.exporter.prometheus component that would set series value incorrectly for stale metrics (@YusifAghalar)

• loki.source.podlogs: Fixed a bug which prevented clustering from working and caused duplicate logs to be sent. The bug only happened when no selector or namespace_selector blocks were specified in the Alloy configuration. (@ptodev) …

Breaking changes

• import.git: The default value for revision has changed from HEAD to main. (@ptodev) It is no longer allowed to set revision to "HEAD", "FETCH_HEAD", "ORIG_HEAD", "MERGE_HEAD", or "CHERRY_PICK_HEAD".

• The Otel update to v0.112.0 has a few breaking changes:

  • [otelcol.processor.deltatocumulative] Change max_streams default value to 9223372036854775807 (max int). https://github.com/open-telemetry/opentelemetry-collector-contrib/pull/35048
  • [otelcol.connector.spanmetrics] Change namespace default value to traces.span.metrics. https://github.com/open-telemetry/opentelemetry-collector-contrib/pull/34485
  • [otelcol.exporter.logging] Removed in favor of the otelcol.exporter.debug. https://github.com/open-telemetry/opentelemetry-collector/issues/11337

Features

• Add support bundle generation via the API endpoint /-/support (@dehaansa)

• Add the function path_join to the stdlib. (@wildum)

• Add pyroscope.receive_http component to receive and forward Pyroscope profiles (@marcsanmi)

• Add support to loki.source.syslog for the RFC3164 format ("BSD syslog"). (@sushain97)

• Add support to loki.source.api to be able to extract the tenant from the HTTP X-Scope-OrgID header (@QuentinBisson)

• (Experimental) Add a loki.secretfilter component to redact secrets from collected logs. …

Bugfixes

• Update windows_exporter from v0.27.2 vo v0.27.3: (@jkroepke)

  • Fixes a bug where scraping Windows service crashes alloy

• Update yet-another-cloudwatch-exporter from v0.60.0 vo v0.61.0: (@morremeyer)

  • Fixes a bug where cloudwatch S3 metrics are reported as 0

• Issue 1687 - otelcol.exporter.awss3 fails to configure (@cydergoth)

  • Fix parsing of the Level configuration attribute in debug_metrics config block
  • Ensure "optional" debug_metrics config block really is optional

• Fixed an issue with loki.process where stage.luhn and stage.timestamp would not apply default configuration settings correctly (@thampiotr)

• Fixed an issue with loki.process where configuration could be reloaded even if there were no changes. (@ptodev, @thampiotr)

• Fix issue where loki.source.kubernetes took into account all labels, instead of specific logs labels. Resulting in duplication. (@mattdurham)

Security fixes

• Add quotes to windows service path to prevent path interception attack. CVE-2024-8975 (@mattdurham)

Breaking changes

• Some debug metrics for otelcol components have changed. (@thampiotr) For example, otelcol.exporter.otlp's exporter_sent_spans_ratio_total metric is now otelcol_exporter_sent_spans_total.

• [otelcol.processor.transform] The functions convert_sum_to_gauge and convert_gauge_to_sum must now be used in the metric context rather than in the datapoint context. https://github.com/open-telemetry/opentelemetry-collector-contrib/issues/34567 (@wildum)

• Upgrade Beyla from 1.7.0 to 1.8.2. A complete list of changes can be found on the Beyla releases page: https://github.com/grafana/beyla/releases. (@wildum) It contains a few breaking changes for the component beyla.ebpf:

  • renamed metric process.cpu.state to cpu.mode
  • renamed metric beyla_build_info to beyla_internal_build_info

Features

• Added Datadog Exporter community component, enabling exporting of otel-formatted Metrics and traces to Datadog. (@polyrain)
• (Experimental) Add an otelcol.processor.interval component to aggregate metrics and periodically forward the latest values to the next component in the pipeline.

Enhancements

• Clustering peer resolution through --cluster.join-addresses flag has been …