Progressive Rollout requires a health assessment to be performed for each resource. See more information in the new "No Downtime" document, including the Sequence diagrams linked from there.
This should be able to return true or false, and should exist for PipelineRollout, MonoVertexRollout, and ISBServiceRollout, and once we're ready to move to having a single NumaflowControllerRollout, for that as well.
We should start with PipelineRollout and MonoVertexRollout. We need to make sure that:
child resource is in "Running" state
child resource doesn't have Failed Child Conditions (note that "progressing" state is sometimes undifferentiated from "failed" state)
"data criticality" measurement is good. This is something reported by Pipeline's Daemon Server (Numaflow team needs to implement for MonoVertex)
Message from the maintainers:
If you wish to see this enhancement implemented please add a 👍 reaction to this issue! We often sort issues this way to know what to prioritize.
Summary
Progressive Rollout requires a health assessment to be performed for each resource. See more information in the new "No Downtime" document, including the Sequence diagrams linked from there.
This should be able to return true or false, and should exist for PipelineRollout, MonoVertexRollout, and ISBServiceRollout, and once we're ready to move to having a single NumaflowControllerRollout, for that as well.
We should start with PipelineRollout and MonoVertexRollout. We need to make sure that:
Message from the maintainers:
If you wish to see this enhancement implemented please add a 👍 reaction to this issue! We often sort issues this way to know what to prioritize.