squbs
squbs copied to clipboard
Add support for a `policy` in Retry stage
We currently prioritize the "ready to be retried" elements over any elements waiting upstream on in1 when joined flow pulls. Some retry scenarios may prefer that upstream elements be treated with a higher priority than the elements currently being retried. So consider the addition of a Policy that would allows clients to configure the Retry stage to control this priority:
- Prioritize retry elements over ones waiting upstream on in1 (the default)
- Prioritize in1 elements waiting over elements currently being retried.
- Balance mode which alternates elements waiting on in1 with ones currently being retried.