Skip to main content

SKE Health Agent

Overview​

The SKE Health Agent is a health check agent to be executed in Kubernetes Destinations where health checks can be executed.

The release manifest is available at:

http://syntasso-enterprise-releases.s3-website.eu-west-2.amazonaws.com/#k8s-health-agent/

Release Notes​

v0.13.0 (2026-09-07)​

Released in Helm Chart version 0.30.0

Features​

  • Agents on multiple Destinations can now check the same resource without overwriting each other's health records. Set the optional destinationName key in the agent's ConfigMap to keep one record per Destination; see Running the agent on multiple Destinations.

Maintenance​

  • Base image updates.

v0.12.2 (2026-09-02)​

Released in Helm Chart version 0.29.0

Bug Fixes​

  • Fixed a zombie-process leak in the SKE Health Agent image. Every health record written to a Git State Store shells out to git, which forks a git remote-https helper; if the parent was killed mid-operation, the orphaned helper reparented to PID 1, which never reaped it — eventually exhausting the pod's PID limit and stopping health reporting entirely. The image now runs under tini as PID 1, so orphaned helpers are reaped correctly. Bucket State Stores are unaffected.

Security Fixes​

  • Base image updates to address security vulnerabilities.

v0.12.1 (2026-08-24)​

Released in Helm Chart version 0.28.0

Security Fixes​

  • Base image updates to address security vulnerabilities.

v0.12.0 (2026-07-29)​

Released in Helm Chart version 0.27.0

Bug Fixes​

  • Health records can now be written to a BucketStateStore from IPv4-only networks. Amazon S3 endpoints were being rewritten to their dual-stack (IPv4/IPv6) variant, which is unreachable from an IPv4-only VPC; the configured endpoint is now used exactly as set.

Maintenance​

  • Base image updates.

v0.11.1 (2026-07-28)​

Released in Helm Chart version 0.26.0

Bug Fixes​

  • Health records can now be written to a BucketStateStore using authMethod: IAM on EKS, where resolving the IAM credentials previously failed.

Maintenance​

  • Base image updates.

v0.11.0 (2026-07-23)​

Released in Helm Chart version 0.25.0

Features​

  • Health Check pods now run with a restricted, non-root pod-level security context by default, and the agent-owned containers are hardened. Workflow containers inherit these defaults and can override them per container where a root image is required. See the Security context section of the Health Definition reference.
  • spec.inputs on a HealthDefinition is now optional.

Bug Fixes​

  • The Health Agent now creates the CronJob (and its ServiceAccount/RBAC) in, and runs the Health Check in, the same namespace as the HealthDefinition, independent of spec.resourceRef.namespace. See the Namespace section of the Health Definition reference and the Health checks guide.

Security Fixes​

  • Dependency and security updates.

v0.10.0 (2026-06-10)​

Released in Helm Chart version 0.24.0

Chores​

  • Chainguard base image SHA updates

v0.9.0 (2026-06-08)​

Security Fixes​

  • Dependencies security updates

v0.8.1 (2026-05-11)​

Security Fixes​

  • Dependencies security updates

v0.8.0 (2026-05-11)​

Security Fixes​

  • Security updates

v0.7.0 (2026-04-29)​

Released in Helm Chart version v0.21.0

Security Fixes​

  • Security updates

v0.6.1 (2026-04-09)​

Released in Helm Chart version v0.20.0

Security Fixes​

  • Security updates

v0.5.4 (2026-03-23)​

Contains dependency updates.

important

Please note that as of v0.5.3, the SKE Health Agent image will be accessible at ghcr.io not registry.syntasso.io.

Please ensure that your Syntasso Registry secret is updated to reflect this. If it is not updated, your cluster will be unable to pull the image and the health-agent pod will fail to start.

v0.5.3 (2026-03-04)​

Released in Helm Chart version v0.18.0

v0.5.2 (2026-02-11)​

Released in Helm Chart version v0.17.0

Bug Fixes​

  • Bump OpenTelemetry dependencies

v0.5.0 (2025-10-29)​

Released in Helm Chart version v0.15.0

Features​

  • Image Pull Secret for the Health Check Adapter image can now be configured via the Agent config map.

v0.4.1 (2025-10-28)​

Released in Helm Chart version v0.14.0

Maintenance​

  • Security updates

v0.4.0 (2025-08-19)​

Released in Helm Chart version v0.13.0

Features​

  • Job history can now be configured via the SUCCESSFUL_JOBS_HISTORY_LIMIT and FAILED_JOBS_HISTORY_LIMIT environment variables. Those can be set in the Deployment manifest.
  • Improve the health agent deployment labels, making it less likely to clash with other deployments in the same namespace.
  • Bumped the Deployment resources limits and requests.

Maintenance​

  • Remove sh from the Health Adapter image.
  • Bump dependencies.

v0.3.3 (2025-06-23)​

Released in Helm Chart version v0.11.0

Bug fixes​

  • Fixes a bug where the healthcheck-adapter was not being built with cross-architecture support

v0.3.2 (2025-06-11)​

Released in Helm Chart version v0.10.0

Maintenance​

  • Security updates

v0.3.0 (2025-04-24)​

Released in Helm Chart version v0.8.0

Bug Fix​

  • Fixes a bug where the agent would error creating cron jobs as the generated name could hit the 52 character limit. The cron job names are now truncated.

Maintenance​

  • Bump dependencies

v0.2.1 (2025-04-16)​

Released in Helm Chart version v0.7.0

Features​

  • Enable the pprof endpoint (8257b01)

Maintenance​

  • Bump dependencies

v0.2.0 (2025-01-15)​

Released in Helm Chart version v0.6.0

Features​

  • Initial release