v0.4.5

🐛 Bug Fixes

• preserve useful characters when truncating K8s label values (#356 by @peterroelants)

📖 Docs

• add IPRally to happy users in README (#351 by @jrosti)

🛠️ Other Improvements

• update dependency astral-sh/uv to v0.11.16 (#352 by @renovate[bot])
• update ghcr.io/astral-sh/uv docker tag to v0.11.8 (#357 by @renovate[bot])
• update softprops/action-gh-release action to v2.6.2 (#358 by @renovate[bot])
• update ghcr.io/astral-sh/uv docker tag to v0.11.6 (#353 by @renovate[bot])
• switch to devenv (#350 by @danielgafni)
• change default branch to main (by @danielgafni)

❤️ New Contributors

• @peterroelants made their first contribution in #356
• @jrosti made their first contribution in #351

v0.4.4

✨ Features

• add create_cluster_if_needed constructor argument to PipesRayJobClient (#324 by @ashutosh1807)
• add submit_job_params argument to PipesKubeRayJobClient.run (#322 by @ashutosh1807)

🐛 Bug Fixes

• add _force_create_local_cluster workaround for broken create_cluster_if_needed (#346 by @ashutosh1807)
• ignore more k8s errors during RayCluster startup (#344 by @danielgafni)

🛠️ Other Improvements

• add a section for other improvements to changelog
• improve Pipes imports and docs references (#343 by @danielgafni)
• upgrade zensical to 0.0.28 (#342 by @danielgafni)
• add minimal Claude Code config (#341 by @danielgafni)

v0.4.3

This release is focused on improving support for authentication and cross-cluster KubeRay workloads.

Note

If you or your company are using dagster-ray in production, please consider adding yourself to the list here :)

✨ Features

• support resolve_hostname hook in PipesKubeRayJobClient (#340 by @danielgafni)
• add resolve_hostname hook to KubeRayResource (#335 by @danielgafni)
• support authOptions in RayClusterSpec (#314 by @danielgafni)
• add support for custom Ray dashboard address with authentication (#315 by @JosefNagelschmidt)
• add missing fields to RayJobSpec (#321 by @danielgafni)

🐛 Bug Fixes

• use serviceName FQDN instead of serviceIP for cross-cluster connectivity (#319 by @pythonmonty)

📖 Docs

• add kaiko to the list of users (#338 by @pythonmonty)
• add Sanas to the list of users (#339 by @ashutosh1807)
• add cross-cluster & authentication docs (#336 by @danielgafni)
• update Pipes docs with better Ray API references (#329 by @danielgafni)
• switch to Zensical (#327 by @danielgafni)

❤️ New Contributors

• @pythonmonty made their first contribution in #319
• @ashutosh1807 made their first contribution in #339

Release v0.4.2

Added

• RayResource now has new lifecycle hook methods for customizing startup messages (and potentially doing something else):
  • on_create
  • on_ready
  • on_connect
  • on_cleanup

Fixes

• fixed duplicated startup log message for LocalRay
• fixed the top-level env_vars Dagster config field not taking effect for KubeRayInteractiveJob

Full Changelog: v0.4.1...v0.4.2

Release v0.4.1

This release includes a few quality of life fixes. It's also tested against the latest versions of the code dependencies: Dagster 1.12, Ray 2.53, and KubeRay 1.5.

Added

• RayCluster's head pod logs are now displayed on startup timeout or failure

Fixes

• the RayCluster cleanup sensor no longer attempts to cleanup clusters with .metadata.ownerReferences field set.

• Fixed race condition with cluster sharing: previously multiple steps running in parallel could create different RayCluster instances at the same time (that were supposed to be shared). dagster-ray now uses Kubernetes Lease-based leader election to coordinate shared cluster creation, which guarantees that only one of the running steps creates the shared RayCluster.

• runtimeEnvYAML now has all strings fully quoted which fixes passing values such as 1e-5 as runtime_env values. Thanks @JosefNagelschmidt!

• ray_address is now optional for RunLauncherConfig. Thanks @cornettew!

• Prevent the RayCluster cleanup sensor from targeting clusters with .metadata.ownerReferences set.

• address config value can now be omitted for ray_executor, making it use Ray's default cluster address resolution. Thanks @cornettew!

New Contributors

• @cornettew made their first contribution in #292
• @JosefNagelschmidt made their first contribution in #297

Full Changelog: v0.4.0...v0.4.1

Release v0.4.0

This release introduces a new feature that is very useful in dev environments: Cluster Sharing. Cluster sharing allows reusing existing RayCluster resources created by previous Dagster steps. It's implemented for KubeRayCluster Dagster resource. This feature enables faster iteration speed and reduced infrastructure costs (at the expense of job isolation). Therefore KubeRayCluster is now recommended over KubeRayInteractiveJob for use in dev environments.

Learn more in Cluster Sharing docs.

Added

• KubeRayCluster.cluster_sharing parameter that controls cluster sharing behavior.
• dagster_ray.kuberay.sensors.cleanup_expired_kuberay_clusters sensor that cleans up expired clusters (both shared and non-shared). Learn more in docs.
• dagster-ray entry now appears in the Dagster libraries list in the web UI.

Changed

• [:bomb: breaking] - removed cleanup_kuberay_clusters_op and other associated definitions in favor of dagster_ray.kuberay.sensors.cleanup_expired_kuberay_clusters sensor that is more flexible.

Full Changelog: v0.3.1...v0.4.0

Release v0.3.1

Added

• A new failure_tolerance_timeout configuration parameter for KubeRayInteractiveJob and KubeRayCluster. It can be set to a positive value to give the cluster some time to transition out of failed state (which can be transient in some scenarios) before raising an error.

Fixes

• ensure both .head.serviceIP and .head.serviceName are set on the RayCluster while waiting for cluster readiness

Full Changelog: v0.3.0...v0.3.1

Release v0.3.0

This release includes massive docs improvements and drops support for Python 3.9

Changes

• [:bomb: breaking] dropped Python 3.9 support (EOL October 2025)
• [internal] most of the general, backend-agnostic code has been moved to dagster_ray.core (top-level imports still work)

Full Changelog: v0.2.1...v0.3.0

Release v0.2.1

Fixes

• Fixed broken wheel on PyPI

Full Changelog: v0.2.0...v0.2.1

Release v0.2.0

Changed

• KubeRayInteractiveJob.deletion_strategy now defaults to DeleteCluster for both successful and failed executions. This is a reasonable default for the use case.
• KubeRayInteractiveJob.ttl_seconds_after_finished now defaults to 600 seconds.
• KubeRayCluster.lifecycle.cleanup now defaults to always
• [:bomb: breaking] RayJob and RayCluster clients and resources Kubernetes init parameters have been renamed to kube_config and kube_context.

Added

• new enable_legacy_debugger configuration parameter to subclasses of RayResource
• new on_exception option for lifecycle.cleanup policy. It's triggered during resource setup/cleanup (including KeyboardInterrupt), but not by user @op/@asset code.
• KubeRayInteractiveJob now respects lifecycle.cleanup. It defaults to on_exception. Users are advised to rely on built-in RayJob cleanup mechanisms, such as ttlSecondsAfterFinished and deletionStrategy.

Fixes

• removed ignore_reinit_error from RayResource init options: it's potentially dangerous, for example in case the user has accidentally connected to another Ray cluster (including local ray) before initializing the resource.