prefect

This release includes:

🔄 Improvements to deploying with flow.from_source

• Integrate storage blocks seamlessly with flow.from_source for synchronizing credentials and configurations. This allows you to update code storage credentials without re-deploying flows.

🔧 New Settings

• We've added a client-side setting for specifying a default work pool.
• Host and port settings for the worker webserver are now customizable.

📚 Updated Guides

• We've updated our work pool-based deployment guide to include information on .deploy.

🧑‍🤝‍🧑 Contributors

• A special thanks to @yarikoptic for their first contribution to Prefect! Also, a big thanks to @taljaards for their continued support and contributions.

See the the release notes for more details!

Release 2.14.4

New improved flow run graph with dependency layout

The flow run graph in the Prefect UI has been rebuilt from the ground up, offering significantly improved performance capabilities that allow larger flow runs to be displayed much more smoothly. We’ve added three new layouts: two non-temporal layout options, designed to provide a clearer picture of the dependency paths, and one to facilitate easy comparison of run durations. The x-axis can now be independently scaled for temporal layouts; and you can adjust it in the graph settings or with the new keyboard shortcuts - and +. We included additional small bug fixes, including the display of cached tasks.

See the following pull requests for implementation details:

• https://github.com/PrefectHQ/prefect/pull/11112
• https://github.com/PrefectHQ/prefect/pull/11105
• https://github.com/PrefectHQ/prefect/pull/11113
• https://github.com/PrefectHQ/prefect/pull/11132
• https://github.com/PrefectHQ/prefect/pull/11138

Enhancements

• Add API route for block counts — https://github.com/PrefectHQ/prefect/pull/11090
• Improved tag handling on DeploymentImage for .deploy:
  • https://github.com/PrefectHQ/prefect/pull/11115
  • https://github.com/PrefectHQ/prefect/pull/11119
• Allow image passed into .deploy to be optional if loading flow from storage — https://github.com/PrefectHQ/prefect/pull/11117
• Ensure client avoids image builds when deploying to managed work pools — https://github.com/PrefectHQ/prefect/pull/11120
• Add SIGTERM handling to runner to gracefully handle timeouts — https://github.com/PrefectHQ/prefect/pull/11133
• Allow tasks to use get_run_logger w/o parent flow run — https://github.com/PrefectHQ/prefect/pull/11129
• Allow ResultFactory creation from_task when no flow_run_context — https://github.com/PrefectHQ/prefect/pull/11134

Fixes

• Avoid printing references to workers when deploying to managed pools — https://github.com/PrefectHQ/prefect/pull/11122

Documentation

• Fix docstring for flow.deploy method example — https://github.com/PrefectHQ/prefect/pull/11108
• Add warning about image architecture to push pool guide — https://github.com/PrefectHQ/prefect/pull/11118
• Move webhooks guide to Development section in guides index — https://github.com/PrefectHQ/prefect/pull/11141

See the release notes for more details!

Deployment Status Monitoring 🚦

• Track deployment status in the Prefect UI
• Emit Prefect Cloud events for deployment status changes, with automation triggers available

Expanded Storage for flow.from_source 🗂

• Load flows from remote storage locations like S3 and Azure Blob Storage using fsspec protocols

Enhancements and Fixes 🔧

• Skip Docker builds with flow.deploy
• Fixes for cancellation cleanup, Pydantic V2 models in flow signatures

Documentation and Guides 📘

• Adds documentation for Systemd daemonization, global concurrency clarification, and Prefect Cloud teams

See the release notes for details!

This release includes the ability to pass **kwargs to state change hooks 🪝, allowing you to further customize the actions you to take upon completion or cancellation. Additionally, we've improved the management of on_cancellation and on_crashed flow run state change hooks.

See the release notes for details!

This release includes:

🪲 Fixes

• Refined flow parameter validation to use the correct validation depending on if the parameter is a pydantic v1 or v2 model.

📚 Documentation Improvements

• Added Python serve and deploy options to the schedules concepts documentation

See the release notes for details!

This release includes:

🌐 prefect-client

This slimmed down version of prefect has a small surface area of functionality and is intended for interacting with the Prefect server or Prefect Cloud only. It is ideal for use in lightweight, ephemeral execution environments when the goal is to interact with Prefect API resources. It is now installable via pip:

pip install prefect-client

🏷 Flow Labels for Subflow Runs
We've improved the Prefect UI by adding flow names to subflow runs.

🪲 Windows Environment Compatibility
We've fixed issues related to pulling flows and building deployments in Windows environments.

🚮 UI Cleanup
We've removed an unnecessary work queue health indicator from push pools.

📚 Documentation Improvements

• Added consistent capitalization
• Updated the Docker guide to mention image builds with prefect.yaml and flow.deploy
• Updated Kubernetes guide with Python-based flow deploy creation methods

For a complete overview of all the changes in this release, visit the compare page.

This release includes:

🚀 flow.deploy
We're excited to introduce flow.deploy as a simple transition from running your served flows on persistent infrastructure to executing your flows on dynamically provisioned infrastructure via work pools and workers. flow.deploy ensures your flows execute consistently across environments by packaging your flow into a Docker image and making that image available to your workers when executing your flow.

See the release notes for details!

This release includes:

⚙️ 🥳 Pydantic V2 Support

In #10946 we add support for Pydantic V2. In 2.13.5 we released experimental support for Pydantic V2, which made it technically co-installable via forced install. In this release we're enabling proper co-installation.

This release includes:

📦 Default result storage block
The new PREFECT_DEFAULT_RESULT_STORAGE_BLOCK setting allows users to set a default storage block on a work pool or via job variables for a deployment. It's now easier to swap result storage locations by updating the environment in the UI or your prefect.yaml file, eliminating the need to alter your flow source code.

🧪 Experimental enhanced cancellation
We're enhancing the reliability of flow run cancellation, which you can try out by enabling the PREFECT_EXPERIMENTAL_ENABLE_ENHANCED_CANCELLATION feature. When enabled, you can cancel flow runs where cancellation can fail, such as when your worker is offline.

See the release notes for details!

This release introduces a few new features:

⚡ You can now load and serve flows from a git repository! With the new flow.from_source method, you can specify a git repository and a path to a flow file in that repository. This method will return a flow object that can be run or served with flow.serve().

🧪 Experimental Pydantic 2 compatibility! We're working eagerly toward having prefect installable with either pydantic<2 or pydantic>2. As a first step toward compatibility, we've ensured that Prefect's use of pydantic is isolated from your use of pydantic in as many ways as possible.

See the release notes for details!

Enhancements

• Lift API and database constraints that require task runs to have an associated flow run id — https://github.com/PrefectHQ/prefect/pull/10816

Fixes

• Fix an issue with infinite scrolling on the sub flow runs tab in the UI - https://github.com/PrefectHQ/prefect-ui-library/pull/1788

Documentation

• Add dark mode base job template screenshot to work pools documentation — https://github.com/PrefectHQ/prefect/pull/10849
• Drop beta tag from push work pools documentation — https://github.com/PrefectHQ/prefect/pull/10799
• Improve logo sizing and general housekeeping - https://github.com/PrefectHQ/prefect/pull/10830

This release includes:

🛠️ CLI Enhancement for Work Pool Configurations
Configure your work pool’s base job template directly through CLI! This enhancement adds more flexibility in creating and updating the base job template of a work pool, providing better control and version management.

👀 Customizable Default Flow Runs View
Optimize your workflow with a tailored default view on your Flow Runs page! A locally stored setting allows a user-centric approach, ensuring your preferred view is a click away.

See the release notes for details!

This release introduces several key features and improvements:

🔒 Server-side enforcement of deployment parameter schemas

🚀 Enhanced deployment flexibility with pattern-based deploying

🛠️ Various fixes, including Prefect variable resolution in the deployments section of prefect.yaml

📚 Documentation updates and 🤝 new contributions from the community

For a detailed breakdown, check out the release notes!

This release introduces several enhancements and fixes:

📊 Option to "Hide subflow runs" on the flow runs page of the Prefect UI.
✨ Multiple enhancements like run_count in prefect.runtime and deployment triggers via CLI.
🛠️ Various fixes including DST handling in CronSchedule and mobile UI improvements.
📚 Documentation updates and 🤝 new contributions from the community.

See the release notes for details!

This release introduces exciting new features, as well as many enhancements, guides, and fixes!

🎛️ Global concurrency limits: Control task execution and system stability with Prefect's new global concurrency and rate limits.

🚦 Work pool and worker status: Work pools and workers can now communicate their status. Ready, Not Ready and Paused for work pools and Online and Offline for workers.

🌌 We deprecated references to orion in our codebase six months ago, in this release, we're removing those references.

📚 Documentation Updates: Updated deployment concept documentation, new guide for deploying Kubernetes to Azure AKS, along with updated screenshots and broken link fixes.

Check out the release notes for more in-depth information and get ready to dive into a more powerful and user-friendly Prefect!

This release includes some important fixes and enhancements. In particular, it resolves an issue preventing the flow run graph from rendering correctly in some cases. Please see the release notes for details.

This release introduces exciting new features, as well as many enhancements, guides, and fixes!

🔥 Introducing Flow.serve(): Deploying flows has never been this straightforward. Just add .serve() to your existing flows and you can now schedule or trigger runs via the Prefect UI and CLI. Deployment and execution are just a script away!

🎨 UI Makeover: The Prefect UI has a fresh new look, with updates in both light and dark modes. Navigate through an interface that's both visually pleasing and more harmonious.

🔧 Enhancements Galore: From improved validation to JSON overrides in deployments and expanded Docker image support, this release has lots of under-the-hood improvements.

👷 Worker and Work Pool Settings: New settings allow more granular control over SQLAlchemy connection pool sizes and better formatting of GitLab and Bitbucket tokens, enhancing the deployment process.

🐞 Fixes: Various bugs have been addressed, including race conditions in concurrent subflow runs, Prefect Cloud login issues, and more to make your Prefect experience smoother.

📚 Documentation Updates: New quickstart guides, detailed Kubernetes usage, and a more organized index make it easier for both newcomers and experienced users to find what they need.

📈 Additional Stats and Insights: Get a snapshot of your flows' performance with new summary graphs and statistics on each flow page.

Check out the release notes for more in-depth information and get ready to dive into a more powerful and user-friendly Prefect!

This release includes two new guides for moving data to and from cloud providers and running flows via Kubernetes along with many enhancements and fixes. Please see the release notes for details.

This release includes a new guide for upgrading from agents to workers, as well as a brand new way to visualize flows without running them. Please see the release notes for details.

This release includes enhanced support for environment variables in deployments, as well as a CLI enhancement and documentation updates. Please see the release notes for details.