Workflow TTL Strategy based on number of workflows in different phases

[terrytangyuan]

Summary

What change needs making?

The current TTL strategy is time-based to delete workflows after x/y/z seconds of success/failure/completion.

Alternatively, we can support only deleting a subset of the workflows. For example, only keep a maximum of x/y/z number of succeeded/failed/completed workflows. All remaining completed workflows are deleted.

Use Cases

When would you use this?

Some users want to keep a number of workflows in each phase for analysis or debugging purposes. It's also easier to see a list of current running workflows + history of the failed/successful ones in one page.

---

Message from the maintainers:

Love this enhancement proposal? Give it a 👍. We prioritise the proposals with the most 👍.

[agilgur5]

Alternatively, we can support only deleting a subset of the workflows. For example, only keep a maximum of x/y/z number of succeeded/failed/completed workflows. All remaining completed workflows are deleted.

This was/is possible already with retentionPolicy as of #6854.
Notably, that wasn't documented at all until #11297, and could still use a reference in the Cost Optimization page