danielgafni · danielgafni · Oct 10, 2025
diff --git a/CHANDELOG.md b/CHANDELOG.md
@@ -13,45 +13,45 @@ Learn more in [Cluster Sharing docs](tutorial/kuberay.md/#cluster-sharing).
 
 ### 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 mode in [docs](api/kuberay.md#dagster_ray.kuberay.sensors.cleanup_expired_kuberay_clusters)
-- `dagster-ray` entry now appears in the Dagster libraries list in the web UI
+- `dagster_ray.kuberay.sensors.cleanup_expired_kuberay_clusters` sensor that cleans up expired clusters (both shared and non-shared). Learn more in [docs](api/kuberay.md#dagster_ray.kuberay.sensors.cleanup_expired_kuberay_clusters).
+- `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
+- [: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.
 
 ## 0.3.1
 
 ### Added
 - `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
+- ensure both `.head.serviceIP` and `.head.serviceName` are set on the `RayCluster` while waiting for cluster readiness.
 
 ## 0.3.0
 
-This release includes massive docs improvements and drops support for Python 3.9
+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)
+- [: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).
 
 ## 0.2.1
 
 ### Fixes
 
-- Fixed broken wheel on PyPI
+- Fixed broken wheel on PyPI.
 
 ## 0.2.0
 
 ### 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`
+- `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
-- `enable_legacy_debugger` configuration parameter to subclasses of `RayResource`
+- `enable_legacy_debugger` configuration parameter to subclasses of `RayResource`.
 - `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`.
 
@@ -64,7 +64,7 @@ This release includes massive docs improvements and drops support for Python 3.9
 - [:bomb: breaking] `RayResource`: top-level `skip_init` and `skip_setup` configuration parameters have been removed. The `lifecycle` field is the new way of configuring steps performed during resource initialization. `KubeRayCluster`'s `skip_cleanup` has been moved to `lifecycle` as well.
 - [:bomb: breaking] injected `dagster.io/run_id` Kubernetes label has been renamed to `dagster/run-id`. Keys starting with `dagster.io/` have been converted to `dagster/` to match how `dagster-k8s` does it.
 - [:bomb: breaking] `dagster_ray.kuberay` Configurations have been unified with KubeRay APIs.
-- `dagster-ray` now populates Kubernetes labels with more values (including some useful Dagster Cloud values such as `git-sha`)
+- `dagster-ray` now populates Kubernetes labels with more values (including some useful Dagster Cloud values such as `git-sha`).
 
 ### Added
 - `KubeRayInteractiveJob` -- a resource that utililizes the new `InteractiveMode` for `RayJob`. It can be used to connect to Ray in Client mode -- like `KubeRayCluster` -- but gives access to `RayJob` features, such as automatic cleanup (`ttlSecondsAfterFinished`), retries (`backoffLimit`) and timeouts (`activeDeadlineSeconds`).

diff --git a/docs/api/kuberay.md b/docs/api/kuberay.md
@@ -132,7 +132,7 @@ These resources initialize Ray client connection with a remote cluster.
 
 A Dagster sensor that monitors shared `RayCluster` resources created by the current Dagster [code location](https://docs.dagster.io/deployment/code-locations/managing-code-locations-with-definitions) (with a `dagster/code-location=<current-code-location>` label selector) and submits jobs to delete clusters either:
 - use [Cluster Sharing](../tutorial/kuberay.md#cluster-sharing) (`dagster/cluster-sharing=true`) and have expired
-- are older than `DAGSTER_RAY_CLUSTER_EXPIRATION_SECONDS` (defaults to 24 hours)
+- are older than `DAGSTER_RAY_CLUSTER_EXPIRATION_SECONDS` (defaults to 4 hours)
 
 By default it monitors the `ray` namespace. This can be configured by setting `DAGSTER_RAY_NAMESPACES` (accepts a comma-separated list of namespaces).
 

diff --git a/docs/tutorial/kuberay.md b/docs/tutorial/kuberay.md
@@ -144,7 +144,7 @@ from dagster_ray.kuberay.configs import RayClusterConfig, ClusterSharing
 
 ray_cluster = KubeRayCluster(
     ray_cluster=RayClusterConfig(
-        cluster_sharing=ClusterSharing(enabled=True, ttl_seconds=3600)
+        cluster_sharing=ClusterSharing(enabled=True, ttl_seconds=30 * 60)
     )
 )
 ```

diff --git a/src/dagster_ray/configs.py b/src/dagster_ray/configs.py
@@ -10,7 +10,7 @@
 DAGSTER_RAY_NAMESPACES_ENV_VAR = "DAGSTER_RAY_NAMESPACES"
 DAGSTER_RAY_NAMESPACES_DEFAULT_VALUE = "ray"
 DAGSTER_RAY_CLUSTER_EXPIRATION_SECONDS_ENV_VAR = "DAGSTER_RAY_CLUSTER_EXPIRATION_SECONDS"
-DAGSTER_RAY_CLUSTER_EXPIRATION_SECONDS_DEFAULT_VALUE = str(24 * 60 * 60)
+DAGSTER_RAY_CLUSTER_EXPIRATION_SECONDS_DEFAULT_VALUE = str(4 * 60 * 60)
 
 
 class Lifecycle(dg.Config):

diff --git a/src/dagster_ray/kuberay/configs.py b/src/dagster_ray/kuberay/configs.py
@@ -309,9 +309,6 @@ class MatchDagsterLabels(dg.Config):
     )
 
 
-DEFAULT_CLUSTER_SHARING_TTL_SECONDS = 60 * 60.0
-
-
 class ClusterSharing(dg.Config):
     """Defines the strategy for sharing `RayCluster` resources with other Dagster steps.
 
@@ -325,6 +322,5 @@ class ClusterSharing(dg.Config):
         default=None, description="Additional user-provided labels to match on."
     )
     ttl_seconds: float = Field(
-        default=DEFAULT_CLUSTER_SHARING_TTL_SECONDS,
         description="Time to live for the lock placed on the `RayCluster` resource, marking it as in use by the current Dagster step.",
     )
diff --git a/src/dagster_ray/kuberay/resources/raycluster.py b/src/dagster_ray/kuberay/resources/raycluster.py
@@ -45,7 +45,7 @@ class KubeRayCluster(BaseKubeRayResource):
     )
 
     cluster_sharing: ClusterSharing = Field(
-        default_factory=ClusterSharing,
+        default_factory=lambda: ClusterSharing(enabled=False, ttl_seconds=15 * 60),
         description="Configuration for sharing the `RayCluster` across Dagster steps. Existing clusters matching this configuration will be reused without recreating them. A `dagster/sharing=true` label will be applied to the `RayCluster`, and a `dagster/lock-<run-id>-<step-id>=<lock>` annotation will be placed on the `RayCluster` to mark it as being used by this step. Cleanup will only proceed if the `RayCluster` is not being used by any other steps, therefore cluster sharing should be used in conjunction with [dagster_ray.kuberay.sensors.cleanup_expired_kuberay_clusters][] sensor.",
     )
 

diff --git a/src/dagster_ray/kuberay/sensors.py b/src/dagster_ray/kuberay/sensors.py
@@ -23,7 +23,7 @@ def cleanup_expired_kuberay_clusters(
 ) -> Generator[dg.RunRequest | dg.SkipReason, None, None]:
     f"""A Dagster sensor that monitors shared `RayCluster` resources created by the current code location and submits jobs to delete clusters that either:
         - use [Cluster Sharing](../tutorial/#cluster-sharing) (`dagster/cluster-sharing=true`) and have expired
-        - are older than `DAGSTER_RAY_CLUSTER_EXPIRATION_SECONDS` (defaults to 24 hours)
+        - are older than `DAGSTER_RAY_CLUSTER_EXPIRATION_SECONDS` (defaults to 4 hours)
 
     By default it monitors the `ray` namespace. This can be configured by setting `{DAGSTER_RAY_NAMESPACES_ENV_VAR}` (accepts a comma-separated list of namespaces)."""
     assert context.code_location_origin is not None

diff --git a/tests/kuberay/test_raycluster.py b/tests/kuberay/test_raycluster.py
@@ -483,7 +483,7 @@ def test_cluster_sharing(
             spec=RayClusterSpec(head_group_spec=head_group_spec, worker_group_specs=worker_group_specs),
         ),
         redis_port=get_random_free_port(),
-        cluster_sharing=ClusterSharing(enabled=True),
+        cluster_sharing=ClusterSharing(enabled=True, ttl_seconds=10 * 60),
     )
 
     @dg.asset