HVR 6.2 raises an F_ error, and keeps it in the UI until someone clears it by hand, even when the next retry recovers on its own. One example is a transient Snowflake PUT upload failure (F_JD024D, driver error 40).

Looking for a way for logged issues that HVR resolved to not be displayed in UI.  

Maybe error notifications configurable:
1. Show all errors (current behaviour).
2. Show an error only if it repeats N times within a time window.
3. Show an error only if it's still unresolved at the end of the cycle or after the retry.

Could mark UI notifications as resolved automatically once the job succeeds again, and apply the same mode to alerts.

Errors that recover on their own are noise. They teach operators to ignore notifications, and real, lasting errors get lost among them.