Daily broken-build defense
Surface every broken solution build the same day, with a ping - and keep the quota clean on its own.
A release/build manager accountable for the daily build's health
The trigger. “Know the moment a solution build breaks, and don’t let dead test instances eat our quota.”
The old way
Escaped defects are found late, and sometimes a customer finds them first. Engineers watch the portal to see what failed. Abandoned instances pile up against the per-value-stream quota until new daily deployments fail because the quota is full. There is no proactive alert, and there is no automatic cleanup.
The Flow way
A scheduled supervisor flow over the registered solution pipelines:
Cron (6pm EST) → Run / collect all registered solution pipelines (fan-out + Merge)
→ Publish to dashboard (Action: portal) → on any fail:
Notify VS broken-build (Google Chat) + Retain instance for RCA; on pass:
Auto-delete instance (quota hygiene). An optional
Check quota before provision guard branch can run before provisioning.
Green runs need no attention. Red runs ping the team and keep their evidence. Passing instances clean themselves up, so the quota never silently blocks the next run.
Payoff
| Old way | Flow way | |
|---|---|---|
| Time to know it broke | Days / found by customer | Same evening, pushed alert |
| Quota | Fills with dead instances | Auto-cleaned on pass |
| Attention needed | Babysit the dashboard | Exceptions only |