NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI · #2178 most downloaded on PyPI
Last release 2 days ago
21 Sep 2026
Ships on a steady schedule
a new release about every 2 weeks
No release notes found
nothing matched a version
10 versions withdrawn
withdrawn after publishing
5 years old
429 releases · first in 2021
One column per quarter.
Nothing published for this version
Fixed an issue where the backfills page would fail to load when a deployment contained backfills from a code location that was no longer in the worksp
Nothing published for this version
Nothing published for this version
Nothing published for this version
You can now set up automatic retries for Jobs. Set a default for all Jobs in your deployment using run_retries.max_retries in Deployment Settings, or
run_retries.max_retries in Deployment Settings, or set it per run using the dagster/max_retries tag. When configured, failed runs will be reexecuted from failure up to the maximum number of times.Alert Policies can now be edited in Dagit. To view and configure the alert policies for your deployment, click on the profile picture in the upper-rig
When using the Kubernetes agent, you can now supply a resources key under workspace that will apply resource requirements to any pods launched by the
resources key under workspace that will apply resource requirements to any pods launched by the agent. For example:workspace:
resources:
limits:
cpu: 100m
memory: 128Mi
requests:
cpu: 100m
memory: 128Mi
You can also vary the resource requirements for pods from different code locations by setting config in the Workspace tab. For example:
location_name: test-location
image: dagster/dagster-cloud-template:latest
code_source:
package_name: dagster_cloud_template
container_context:
k8s:
resources:
limits:
cpu: 100m
memory: 128Mi
requests:
cpu: 100m
memory: 128Mi
dagster-cloud CLI to add or update a code location, CLI options can now override location config from a specified file. For example, dagster-cloud workspace add-location --from-file location.yaml --image dagster/dagster-cloud-examples:64c9c2.Fixed an issue where the Launchpad in Dagit sometimes incorrectly launched in an empty state.
Fixed an issue where Dagit sometimes showed a generic "Internal Server Error" message when launching a run, instead of a more detailed error message.
Configuring tags for an alert policy is now optional. When omitted, the alert policy will apply to all runs in the deployment that have matching event
dagster-cloud CLI using dagster-cloud workspace pull.You can now set a container_context key in your code locations in the Workspace tab, which lets you set configuration for a specific execution environ
container_context key in your code locations in the Workspace tab, which lets you set configuration for a specific execution environment (K8s, Docker, or ECS) for that code location. Previously, this configuration could only be set in the dagster.yaml file for your agent, which required you to share it across code locations, and restart your agent whenever it changed. For example, you can specify that a code location should include a secret called my_secret and run in a K8s namespace called my_namespace whenever the Kubernetes agent creates a pod for that location:location_name: test-location
image: dagster/dagster-cloud-template:latest
code_source:
package_name: dagster_cloud_template
container_context:
k8s:
namespace: my_namespace
env_secrets:
- my_secret
For more information, see https://docs.dagster.cloud/guides/adding-code#environment-specific-config.
Nothing published for this version
Dagster Cloud Slack notifications now display message previews.
Nothing published for this version
Email alerts now include the run’s start and end time.
Nothing published for this version
Added the ability to add or update single code locations with a YAML file in the dagster-cloud CLI. For more information, see the adding code to Dagst
Added the ability to add or update single code locations with a YAML file in the dagster-cloud CLI. For more information, see the adding code to Dagster Cloud docs (https://docs.dagster.cloud/guides/adding-code#using-the-dagster-cloud-cli).
When Dagster Cloud is temporarily unavailable due to scheduled maintenance, jobs that are running when the maintenance starts will wait for the maintenance to conclude and then continue execution, instead of failing.
Alert policies are now supported in Dagster Cloud. Alert policies define which jobs will trigger an alert, the conditions under which an alert will be sent, and how the alert will be sent.
An alert policy includes a set of configured tags. Only jobs that contain all the tags for a given alert policy are eligible for that alert. Additionally, an alert policy includes a set of conditions under which an alert will be triggered. For example, the alert can be triggered to fire on job failure, job success, or both events. Finally, an alert policy specifies where the alert will be sent. Currently, we support Slack and email as targets for an alert.
See https://docs.dagster.cloud/guides/alerts for more details.
Nothing published for this version
Nothing published for this version
Nothing published for this version
You can now set secrets with the ECS Agent using the same syntax that you use to set secrets in the ECS API>).
Bugfixes
dagster_cloud_examples package to fail to load when used with the local agent.Documentation
Nothing published for this version
[K8s] The agent now monitors active runs. Failures in the underlying Kubernetes Job (e.g. an out of memory error) will now be reported in Dagit.
Nothing published for this version
Individual context.log messages which appear in the Dagster event log will now be truncated after 50,000 characters. The full contents of these messag
context.log messages which appear in the Dagster event log will now be truncated after 50,000 characters. The full contents of these messages remain available in the compute logs tab. For large logs, we recommend logging straight to stdout or stderr rather than using context.log.Nothing published for this version
Sensors that have a default status can now be manually started. Previously, this would fail with an invariant exception.
Nothing published for this version
Added a button to test out Dagster Cloud with a sample code location when you go to the Workspace tab on an empty Deployment.
imagePullGracePeriod field to the helm chart that tells the agent how long to allow errors while pulling an image before failing. For example:# values.yaml
workspace:
imagePullGracePeriod: 60
docker.io/dagster/dagster-cloud-agent image includes a dagster user with ID 1001 that can be assumed by setting podSecurityContext in your Helm values:# values.yaml
podSecurityContext:
runAsUser: 1001
Nothing published for this version
The Dagster Cloud workspace page now allows creating, deleting, and updating code locations from Cloud Dagit in addition to via the CLI.
secrets_tag parameter to None, which will cause it to not look for any secrets to be included in the tasks that the agent creates. This can be useful in situations where the agent does not have permissions to query AWS Secretsmanagerdagit_url property to the DagsterInstance in Dagster Cloud that can be used to reference the Dagster Cloud Dagit URL within ops and sensors.dagster.yaml, by using CLI arguments.pythonLogs:
# The names of python loggers that will be captured as Dagster logs
managedPythonLoggers:
- foo_logger
# The log level for the instance. Logs emitted below this severity will be ignored.
# One of [NOTSET, DEBUG, INFO, WARNING, WARN, ERROR, FATAL, CRITICAL]
pythonLogLevel: INFO
# Python log handlers that will be applied to all Dagster logs
dagsterHandlerConfig:
handlers:
myHandler:
class: logging.FileHandler
filename: "/logs/my_dagster_logs.log"
mode: "a"
Nothing published for this version
Secrets in the ECS Agent: When using Dagster Cloud in ECS, you can specify a list of AWS Secrets Manager ARNs to include in all tasks that the agent s
dagster.yaml file as follows:user_code_launcher:
module: dagster_cloud.workspace.ecs
class: EcsUserCodeLauncher
config:
cluster: your-cluster-name
subnets:
- your-subnet-name
service_discovery_namespace_id: your-service-discovery-namespace-id
execution_role_arn: your-execution-role-arn
log_group: your-log-group
secrets_tag: "my-tag-name"
secrets:- "arn:aws:secretsmanager:us-east-1:1234567890:secret:MY_SECRET"`
Values.computeLogs.enabled to false, users can prevent compute logs from being forwarded into Dagster Cloud. Logging configured separately from Dagster on Kubernetes will continue to work, but won’t be viewable in the Dagster Cloud UI.dagster.yaml.Nothing published for this version
Dagster Cloud now supports authenticating with Github. Verified emails associated with your Github user can be used for login authorization for Dagste
dagster-cloud --version.-h as another alias for --helpdagster-cloud config setup, your default values will now show in the help text for any command.--install-completion no longer requires you to pass the name of your shell.Nothing published for this version
dagster-cloud config setup now allows the user to authenticate the CLI by logging in through the browser.
dagster-cloud config setup now allows the user to authenticate the CLI by logging in through the browser.dagster-cloud CLI, a deployment no longer needs to be specified.If your agent’s Dagster Cloud version is >=0.13.15, its version will now surface on your Dagster Cloud instance status page.
The Kubernetes agent can now specify a dictionary of labels that will be applied to the pods spun up by the agent. Here is an example values.yaml snippet that adds pod labels:
workspace:
labels:
my_label_key: my_label_value
Nothing published for this version
Nothing published for this version
Nothing published for this version
Deployment settings (run queue configuration, run start timeouts) can now be configured via the dagster-cloud CLI. For example: dagster-cloud deployme
Deployment settings (run queue configuration, run start timeouts) can now be configured via the dagster-cloud CLI. For example: dagster-cloud deployment settings set-from-file example-settings.yaml. These settings will soon be available to configure within Dagit as well as the CLI.
# example-settings.yaml
run_queue:
max_concurrent_runs: 10
tag_concurrency_limits: []
run_monitoring:
start_timeout_seconds: 300
There is now documentation to set up SAML SSO for Google Workspace. See https://docs.dagster.cloud/auth/google-workspace for details.
Containers created by the Docker agent can now be configured with any configuration that’s available on the Container.run call in the Docker Python client. For example, to ensure that containers are automatically removed when they are finished running, you can configure your dagster.yaml as follows:
user_code_launcher:
module: dagster_cloud.workspace.docker
class: DockerUserCodeLauncher
config:
container_kwargs:
auto_remove: true
Logging is now emitted from the dagster_cloud logger.
Nothing published for this version
Performance improvements in Dagit on the Run Viewer page, Runs page, and Workspace page.
dagster-cloud CLI will now notify the user if their Dagster Cloud agent is not running when attempting to modify workspace locations, instead of polling for the agent to sync the update.dagster-cloud configure CLI command has been renamed to dagster-cloud config setup.dagster-cloud CLI.dagster-cloud CLI configuration process.dagster_cloud_api.agent_label configuration option in the dagster.yaml file.Nothing published for this version
[ECS] Previously, ECS tasks and services created by the ECS agent always used the VPC’s default security group. Now, you can configure it to use a dif
dagster.yaml.dagster-cloud workspace add-location --help to see the full set of available options.Nothing published for this version
Email alerts now display basic metadata about the failed run, including the number of steps succeeded, steps failed, and a link to the run in your Dag
dagster-cloud CLI now correctly exits on keyboard interrupt.Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Your coding agent can read these notes before it upgrades. Set up the MCP server →