Argo Workflows v2.7
What are the big new features?
Submittable WorkflowTemplates
Treat a workflow template as a library component and submit this directly from the UI and CLI.
Prometheus Metrics
Emit custom workflow-level and template- level metrics to Prometheus. This could be useful to know
- workflow/step durations
- step failure counters
- metrics derived from templates (e.g. model training score, error rate, etc.),
- any custom metric that could be useful
to a live dashboard!
Default workflow spec that allow you to configure defaults for all workflows.
This is useful if you want to have something happen for every workflow, e.g. always send a Slack message on error.
Enhanced graph view in the UI
The view can now show 10x as much information making it much easier to figure out why very large (500+ node) workflows failed.
Automatically create and and delete pod-disruption budgets
You may find that workflows are impacted because to make workflows run more reliably by preventing pod from being deleted.
New argo stop command
Gracefully stop workflows.
Support for Alibaba Cloud OSS and Google Cloud and dozens of small enhancements and bug fixes!
Release Notes
- https://github.com/argoproj/argo/releases/tag/v2.7.0-rc1
- https://github.com/argoproj/argo/releases/tag/v2.7.0-rc2
- https://github.com/argoproj/argo/releases/tag/v2.7.0-rc3
