Member-only story

Prefect 2.6 Adds Configurable Results, Dynamic Work Queue Patterns, and Custom Failure Handling

And many more enhancements to the UI, CLI, and docs

Anna Geller
9 min readOct 13, 2022
Prefect as a bot observing the dataflow universe, wondering what you’ll build with these new features

Prefect 2.6 ships with highly configurable results supporting any remote storage block as a backend. This release also adds several highly impactful enhancements, including dynamic work-queue matching, which enables better UX for setting per-deployment concurrency limits, and UI improvements providing more visibility into active runs. After upgrading to 2.6, you’ll be able to trigger parametrized runs from deployments from the terminal and pass failed runs to downstream tasks using the new allow_failure annotation. We’ll cover the enhancements in this post. As always, to learn more, check the release notes and our documentation.

Table of contents· Results
What is a result configuration?
Why are Results important?
· Dynamic work queue patterns
Problem
Solution
Applying this matching pattern to singleton deployments
· Show-Active-Runs button on each work queue page
· New allow_failure annotation to allow failed runs to be passed downstream
· Run parametrized deployments from CLI
· Further improvements
· Next steps

Results

--

--

Anna Geller
Anna Geller

Written by Anna Geller

Data Engineering, AWS Cloud, Serverless & .py. Get my articles via email https://annageller.medium.com/subscribe YouTube: https://www.youtube.com/@anna__geller

No responses yet