Skip to content

amazon-cloudwatch-observability-3.0.0

Compare
Choose a tag to compare

🛑 Changes to defaults 🛑

  • The CloudWatchAgent and FluentBit daemonsets by default now have their PriorityClass set to system-node-critical. Opt-out available via overrides / advanced config (#137)

🚀 New features 🚀

  • Introduce support to manage multiple installations of the AmazonCloudWatchAgent custom resource each of which can now either be a deployment, statefulset or daemonset. Default behavior remains unchanged wherein a single daemonset will be installed (#126)
  • Introduce support for Prometheus scraping wherein a scrape config can now be passed in via agent -> prometheus -> config. Also introduces support for Prometheus Target Allocator to facilitate scalable prometheus scraping via agent -> prometheus -> targetAllocator (#126)

💡 Enhancements 💡

  • Unify schema for Application Signals metrics across envs / platforms
  • Increment ADOT JAVA SDK to v1.32.6 for Application Signals
  • Support p5en instance types for GPU Enhanced Container Insights (#150)

🧰 Bug fixes 🧰

  • Default tolerations are no longer applied to the controller manager deployment and now only apply to the daemonsets (#137)
  • Avoid excessive logging of 404 errors in the CloudWatchAgent pod logs when IMDS tags access is not enabled
  • Fix an edge case that occassinally causes the CloudWatchAgent pod to crash with a concurrent map read/write error