Plan Make incomplete-execution storage before errors pile up
Decide whether failed data must be retained, then monitor the storage used for incomplete executions.
Make documents that stored failures count toward plan storage and that a discard-on-full setting can lose data that cannot be recovered.
Work through the setup
- Classify the workflow. Write whether losing one event is acceptable. A cosmetic notification and a customer order are not equivalent. Keep the business consequence next to the scenario name.
- Review retention controls. Read Store incomplete executions and the setting controlling behavior when storage is full. Do not enable discarding because it makes the dashboard look quieter.
- Set an operational review. Decide who checks unresolved/pending executions and how that person notices a backlog. A manually recorded weekly check is not sufficient for time-critical business work.
- Practice recovery on test data. Use a controlled failure and verify what is retained, what pauses and how you resolve it. Use non-sensitive payloads where possible and follow your own data policy.
A worked example
A workflow receives 100 events a day and a destination outage lasts two days. The useful question is not only “does retry exist?” but “can we retain and inspect the affected events, and will someone notice before the queue fills?” Actual capacity depends on payload size and the current plan; do not assume these example counts fit.
What to watch out for
- Plan storage is not unlimited.
- Discarding data can make the scenario continue while losing events.
- Execution logs may contain personal information.
Reserve room for a realistic failure
Estimate the size and frequency of failed executions from a permitted sample, then compare with the storage allowance shown for the organization. A small number of large payloads can matter more than many tiny rows. Keep an owner and review interval for unresolved executions so the store is not treated as permanent archival storage. Do not assume every new failure will be retained when the allowance is exhausted; read the scenario’s full-storage behavior.
Sources and evidence
- Scenario settings - Help Center (read November 7, 2026)
- Manage incomplete executions - Help Center (read November 7, 2026)
Common checks
Is incomplete-execution storage the same as a permanent backup?
No. It is an operational recovery feature with limits and lifecycle behavior. Keep any required system of record separately.
Published November 7, 2026. Sources last read November 7, 2026. Send a correction.