v1.300066.1

💡Enhancements💡

• Downgrade PMD error logs for systemmetrics pipeline by @sky333999 in #2073

Full Changelog: v1.300066.0...v1.300066.1

v1.300066.0

Released On

🚀New Features 🚀

• Add OTLP export support with receivers, endpoint validation, and agent health integration by @JayPolanco in #2062

💡Enhancements💡

• Add awsneuron metrics processor for Neuron Monitor by @petruanica in #2050
• Add otel components hyperpod efa kubelet by @mitali-salvi in #2066
• Remove ec2tagger from systemmetrics pipeline if IMDS is not accessible by @sky333999 in #2067

Full Changelog: v1.300065.0...v1.300066.0

v1.300065.0

Released On

🐞Bug Fixes🐞

• [Agent] Handle regex escaping for '+' in timestamp format by @jefchien in #2008

💡Enhancements💡

• [Agent] Add systemmetrics pipeline for recognized hosts by @sky333999 in #2052
• [Agent] Adds append_dimensions to collectd, supporting custom metric dimensions by @TravisStark in #1997

Full Changelog: v1.300064.1...v1.300065.0

v1.300064.1

Released On

💡Enhancements💡

• Built with newer version of Go
  • 1.25.6 (S3, SSM, ECR, DockerHub)
  • 1.24.13 (Amazon Linux YUM)

Full Changelog: v1.300064.0...v1.300064.1

v1.300064.0

Released On

💡Enhancements💡

• Bump Expr to v1.17.7

Full Changelog: v1.300063.0...v1.300064.0

v1.300063.0

Released On

New Features

🚀 [ContainerInsights] Add Instance Store NVME metrics translation for EKS environments
Added support for instance store NVME metrics collection in Container Insights for EKS clusters by leveraging the awscontainerinsightsreceiver from the OpenTelemetry contrib package. The agent now automatically scrapes and translates NVME device metrics with proper naming conventions and dimension sets, exported in EMF log format. This provides visibility into local ephemeral storage performance for instance types with NVMe instance stores (I3, I4i, C5d, M5d series), enabling customers to monitor disk I/O, throughput, and utilization for workloads using high-performance local storage.

🚀 [ContainerInsights] Add EFA Metrics to Container Insights for EKS environments
Introduced Elastic Fabric Adapter (EFA) metrics collection for Container Insights on EKS clusters. The agent now collects five critical EFA metrics from the Linux sysfs interface: unresponsive_remote_events, impaired_remote_conn_events, retrans_timeout_events, retrans_pkts, and retrans_bytes. Metrics are aggregated at node, pod, and container levels, enabling customers to detect connectivity issues, identify packet retransmission patterns, and monitor network performance for HPC, ML training, and distributed workloads requiring high-performance inter-node communications.

Bug Fixes

🐞 [Agent/SSM] Fix ManageAgent SSM document compatibility across all supported operating systems
Resolved a critical issue with the ManageAgent SSM document that caused failures on Debian and Ubuntu systems. The v1.300062.0 security fix used bash-specific conditional syntax ([[ ]]) that is incompatible with dash, the default shell (/bin/sh) on Debian-based distributions. The document has been updated to use POSIX-compliant syntax that works across all supported operating systems and their default shells, maintaining the security improvements while ensuring compatibility with Amazon Linux, Debian, Ubuntu, and other supported distributions.

Full Changelog: v1.300062.0...v1.300063.0

v1.300062.0

Released On

🐞Bug Fixes🐞

• [OTLP] Fix OTLP pipeline configuration when used with EKS/ContainerInsights to prevent blocking custom metrics by @jefchien in #1934
• [Agent] Fix postinst failure for Debian package when init system is not available in Docker or chroot environments by @movence in #1929

🚀New Features 🚀

• [ContainerInsights] Enable opt-in for high frequency GPU metrics collection with configurable intervals for ML inference workloads by @yanhaoluo666 in #1893

💡Enhancements💡

• [Logs] Fallback to iterative DescribeLogGroups if batch API not supported in region by @duhminick in #1915
• [ContainerInsights] Add enhanced CI checks and nil checks for neuron decorators to prevent panics by @sky333999 in #1935

New Contributors

• @yanhaoluo666 made their first contribution in #1893

Full Changelog: v1.300061.0...v1.300062.0

v1.300061.0

Released On

🐞 Bug Fixes 🐞

• [Configuration] Append logic for merging JVM/OTLP configurations fix
  The CWA was unable to correctly append JVM/OTLP configurations when using the config (append) action. This change addresses this issue by allowing customers to append multiple configurations of this type. @lisguo #1870

• [Configuration] Add support for EC2 metadata as dimensions
  Fixed an issue where the agent was unable to use the ${aws: formatting to substitute dimensions with EC2 metadata information or tags. @lisguo #1876

• [Configuration] Add support for EC2 tags as dimensions for host metrics
  Fixed an issue where the agent only supports raw key/value pairs for the append_dimensions section of the host metrics config. This fix allows the agent to support dynamic AutoScalingGroup dimension using EC2 metadata. @lisguo #1892

• [Logs] Memory leak fix when using publish_multi_logs flag
  This bug fix addressing a memory leak issue when using the publish_multi_logs flag with dynamically determined log destinations in CloudWatch Logs. This was due to the agent not correctly closing open goroutines for associated log files. @dricross #1848

• [Contrib] Fixed messaging system logic
  Bug fix to facilitate the proper linking of cross service traces @sky333999 amazon-contributing/opentelemetry-collector-contrib#368

• [Agent] Windows startup issue fix
  There was an error in the Agent where it would fail to start on windows on stderr. This change updates the agent control scripts to address the issue. @movence #1906

💡 Enhancements 💡

• [Packaging] Updated OTeL JMX jar used by the agent
  The CWA was using a custom build of the OpenTelemetry JMX Metrics Gatherer JAR. There has been several newer versions released upstream that contain bug fixes for JMX metrics etc. This change removes the custom build in favor of using the upstream version. This also reduced the size of the agent by 200MB.

• [Agent] IPv6 Support on EC2
  Adds CloudWatch Agent support for IPv6 on EC2 INSTANCES @Paramadon #1871

Full Changelog: v1.300060.0...v1.300061.0

v1.300060.0

Released On

🐞 Bug Fixes 🐞

• [Logs] Fixed detection of truncated log files to prevent missing or duplicate log entries
  • Fix state file truncation detection by @jefchien in #1858
• [Prometheus] Fixed Prometheus metrics being sent to undefined log streams in EKS when configured alongside Kubernetes/ECS
  • [Bugfix] EMF exporter not handling prometheus config in container envs (EKS/ECS) correctly by @movence in #1844
• [Processors] Fixed transform processor missing 'error mode' that cause silent failures during metric transformations
  • Fixing CWA error_mode bug in addon v4.2.0 by @Paramadon in #1837
• [OTLP] Fixed receiver port conflict issues
  • Introduce shared receiver model for OTLP to avoid port conflicts by @movence in #1821
• [ContainerInsights] Fixed duplicate metrics collection from Kueue controller by limiting to single collector pod
  • Update contrib components to v0.0.0-20250916160455-365e3e3baf3a by @sky333999 in #1865
  • Update contrib components to v0.0.0-20250916235509-415059d978ca by @sky333999 in #1866
• [Debian] Debian package restarts amazon-cloudwatch-agent after upgrade
  • Restart agent on debian package upgrade by @dricross in #1826

🚀 New Features 🚀

• [ContainerInsights] Added Ingress, PersistentVolume and PersistentVolumeClaim metrics
  • PVs, PVCs and Ingress metrics by @aadam19 in #1846
• [Metrics] Added support for exponential histograms
  • Add exponential histogram support to CloudWatch PMD Exporter by @dricross in #1677

💡 Enhancements 💡

• [Agent] RPM size reduction by optimizing helper binaries
  • Refactor helper binaries to save 161MB of disk space when the agent is installed and reduce RPM by 48MB by @chadpatel in #1454

New Contributors

• @aadam19 made their first contribution in #1846

Full Changelog: v1.300059.0...v1.300060.0

v1.300059.0

Released On

🐞Bug Fixes🐞

• [ApplicationSignals] Fix race conditions in applicationsignals processor by @bjrara in #1758
• [Logs] Update state when batch dropped by @duhminick in #1789
• [Logs] Revert "Add default log publishing concurrency (#1770)" by @jefchien in #1819
• [JMX] Fix infinite loop bug when java is missing by @dricross in amazon-contributing/opentelemetry-collector-contrib#343
• [JMX] Fix permissions check for non-linux users by @lisguo in amazon-contributing/opentelemetry-collector-contrib#344

🚀New Features 🚀

• [Logs/Windows] Add windows event id filtering to CWAgent by @Paamicky in #1737
• [Logs/Windows] Add windows event regex filtering to CWAgent by @Paamicky in #1764
• [Logs/Windows] Add event ID and regex features to CWAgent configuration wizard by @Paamicky in #1774
• [Metrics/Memory] Add mem_shared metric to CloudWatch Agent by @ola-amr in #1792
• [Metrics/NVME] Adding NVME Instance Store Metrics by @Paramadon in #1810
• [ContainerInsights] Add UltraServer support for CloudWatch agent by @petruanica in #1571
• [ContainerInsights] Add new Tensor Core and GPU capacity metrics by @tudor-manea in #1814
• [ContainerInsights] Add Neuron Core and EFA capacity metrics by @sky333999 in #1820

💡Enhancements💡

• [Logs] Zero-allocation optimization for sendLine and readLine by @duhminick in #1803
• [Logs] Dynamic reader buffer for tail by @duhminick in #1809
• [ApplicationSignals] Add aws.remote.resource.account.* attributes to Application Signals EMF logs by @bjrara in #1791

New Contributors

• @Paamicky made their first contribution in #1737
• @ola-amr made their first contribution in #1792
• @petruanica made their first contribution in #1571
• @tudor-manea made their first contribution in #1814

Full Changelog: v1.300058.0...v1.300059.0