Context / Goal
Currently, the system allows for multiple reconciliations for a given dataset id to run at the same time. By allowing only one reconciliation for each dataset to run at a time, we mitigate potential issues such as unintended repeat triggering of reconciliation runs.
Expected Outcome
Suggestions:
- We could track in-memory state somewhere (RecRunService probably).
- When a duplicate rec run is requested, we could respond with
409 CONFLICT or 429 TOO MANY REQUESTS or something like that.
Additional context / implementation notes