github.com/aws/amazon-cloudwatch-agent

Enhancements:

• [Logs] Attach Account ID to entity for cross-account checks

Bug Fixes:

• [Logs] Fix log stream name placeholder resolution for EMF logs
• [Logs] Fix race condition in service provider component of entitystore extension
• [Metrics/Logs] Use IAM credential endpoint from IMDS for accurate entity IAM role name

========================================================================

Enhancements:

• [Metrics/Logs] Annotate Infra, statsD, collectD metrics and application logs with entities
• [ApplicationSignals] Support runtime metrics for Application Signals
• [ApplicationSignals] Create a server to expose service names for instrumented applications on K8s

========================================================================

Enhancements:

• [ContainerInsights] Use LocalNode decorator for DCGM and Neuron metrics

========================================================================

Bug Fixes:

• Fix logs-only check in agent and container environment merge

========================================================================

Features:

• Support OTLP receiver for metrics
• Support AMP as a metrics destination
• Support for supplemental configuration
• Support JMX metric collection for JVM, Tomcat, and Kafka on EKS
• Support JMX Container Insights

Enhancements:

• [Traces] Support OTLP format in transit
• [ApplicationSignals] Update metric processing
• [ContainerInsights] Use protobuf content type for k8s client

Bug Fixes:

• [Metrics/Procstat] Fix memory_swap metrics on Linux

========================================================================

Enhancements:

• Increase retry delay on PLE for certain exceptions
• [ApplicationSignals] Add new log attribute for AWS Resource CloudFormation Primary Identifier in EMF logs
• [ApplicationSignals] Add ECS as a supported platform with ECS specific attributes for EMF logs

========================================================================

Bug Fixes:

• [ApplicationSignals] EKS resource detector is now no-op on non-EKS environments
• Stop converting nanosecond metrics to microsecond

Enhancements:

• [ContainerInsights] Use IMDS to get instance ID and type for Neuron Monitor.

========================================================================

Bug Fixes:

• [ApplicationSignals] Fix EKS detector nil panic (#1315)

Enhancements:

• [ContainerInsights] Add HyperPod status metrics for SageMaker (#1309)
• [ContainerInsights] Add pod_cpu_usage_total & pod_memory_working_set (#1294)
• Reduce the size of the start-amazon-cloudwatch-agent binary (#1302)
• [ContainerInsights] Update the prefix for SageMaker HyperPod status metrics (#1313)

========================================================================

Bug Fixes:

• [ContainerInsights] Update GPU usage metrics emitted
• [ContainerInsights] Deprecate runtime tag from neuron metrics to fix false average calculation

Enhancements:

• [ContainerInsights] Filter some neuron attributes to reduce the size of EMF logs

========================================================================

Bug Fixes:

• [Metrics/Plugin] Drop original metrics support added for statsd, collectd and ethtool metrics
• [Logs/Windows Event] Add windows event log service restart detection and resubscribe
• [Metrics/JMX, Metrics/Net, Metrics/DiskIO] Change cumulative to delta conversion to drop initial value
• [Metrics/JMX] Suppress sessions unit warning

Enhancements:

• [Metrics/JMX] Add cumulative to delta conversion for JMX metrics
• [ApplicationSignals] Rename and group application signals attributes
• Upgrade OTEL Contrib to v0.103.0

========================================================================

Bug Fixes:

• [Container Insights/Prometheus] Fix Prometheus file-path start-up issue in Containers

========================================================================

Enhancements:

• [Application Signals] Optimize memory utilization on Kubernetes
• [Application Signals] Add remote db user attribute to traces and logs
• [Metrics] Support append_dimension for ethtool

Bug Fixes:

• [Metrics/Plugin] Reduce logging for known non-convertible metric units
• [Metrics/Plugin] Fix tomcat.sessions metric to aggregate for all Tomcat applications
• [Metrics/Plugin] Update default collection interval to 60 seconds
• [ApplicationSignals] Add log group names as attribute (#1220)

========================================================================

Features:

• Support JMX metric collection for JVM, Tomcat, and Kafka on EC2

Enhancements:

• [ContainerInsights] Add NVIDIA GPU count metrics
• [Application Signals] Enable detailed logging for metrics/traces when debug is enabled

Bug fixes:

• [Logs/Windows Event] Fix load state offset parsing to support unsigned int
• [ContainerInsights] Prevent non-active workload GPU pod metrics from being emitted

========================================================================

Enhancements:

• [Application Signals] Export EMF logs to /aws/application-signals/data
• [Application Signals] Rename metric namespace to Application Signals
• [Application Signals] Change metric schema from HostedIn to Environment
• Trim AWS prefix for spans in X-Ray exporter

Bug fixes:

• Fix panic when using amazon-cloudwatch-agent-ctl -a cond-restart

========================================================================

Features:

• [Metrics] Append Dimension Volume Id For Metrics

Enhancements:

• Upgrade OTEL Contrib To v0.98.0

========================================================================

Bug fixes:

• Fix nil referencing issue while decorating container insights metrics

========================================================================

Features:

• [ContainerInsights] Add Elastic Fabric Adapter (EFA) observability with Kubernetes

Enhancements:

• Retain CloudWatch Log Group when translating X-Ray segments

Bug fixes:

• Clean up log pusher state file on startup
• Drop container/pod level Nvidia GPU metrics when there is no active workload

========================================================================

Features:

• [ContainerInsights] Add AWS Trainium & Inferentia observability with Kubernetes

========================================================================

Features:

• [ContainerInsights] Add Container Insights for Windows in kubernetes.

========================================================================

Bug fixes:

• [AppSignals] Explicitly set the default GC interval for metrics limiter

Enhancements:

• [ContainerInsights] Add InstanceType attribute to GPU metrics

========================================================================

Features:

• [ContainerInsights] Add NVIDIA GPU observability with Kubernetes

Enhancements:

• [AppSignals] Improve Kubernetes support

========================================================================

Enchancements:

• [AppSignals] Log correlation
• [AppSignals] New Metric Rollup
• [AppSignals] Add metrics cardinality control
• [AppSignals] Add application log group info

========================================================================