Changelog

• Improved retries of select errors on startup of a container task
• The task-agent binaries and pidfile are now installed at more conventional locations for both machine and container runner
• GA release of machine runner 3.0

Machine runner 3.0 is now Generally Available, providing improvements to usability, setup, and maintenance of machine runners.

Keep reading

• Improved the retry logic on task pod startup to prevent errors.
• Release of the Linux machine runner 3.0 preview. For more information see, Installing Machine Runner 3.0 on Linux.

• Reduce claiming interval from 10 seconds to 5 seconds
• Improve robustness when starting tasks on newly created nodes
• Default to a non-root user in the container runner Docker image

What’s New

• CircleCI built in environment variables set as labels on task pods
• Task pod garbage collection improvements
• Improve task-agent task pod injection reliability
• Improve task-agent download reliability

What’s New

• Add circleci symlink to the circleci-agent binary within tasks
• Retry loading resource class token secrets on retryable errors
• Update various dependencies

What’s New

• Upgrade to Go 1.20
• Migrate Helm chart to new image repository: https://hub.docker.com/r/circleci/runner-agent/
• Improve error handling and increase reliability in the task worker loop
• Sample low-value health check logs when using the informational logging level

What’s New

• Bugfixes and package updates

What’s New

• Add lifecycle pre-stop command to gracefully terminate task agent on shutdown
• Ensure all Task lifecycle step events can be sent, even on shutdown of the agent and the context is cancelled
• Don’t delete the task pod if task agent is still running
  • This is related to issue https://github.com/kubernetes/client-go/issues/724. The remotecommand package may not always handle network disconnects on the exec command, returning a nil error. Therefore, we don’t delete the task pod in this case, giving a chance for the task to complete in an “orphaned” state.
• Prevent Task lifecycle step output from being lost during long-running jobs
• Improve task cleanup liveness check robustness and add extra logging
• Fix draining and cleanup of a task on shutdown of the runner agent
• Use stricter task pod config parsing with more helpful error messaging

What’s New

• Various dependency updates
• Fix for task liveness probe retries not being respected
• Fix for task pods not being cleaned up when exiting with a run error