Errors and Recovery

A watch condition can hold more than one error reason at the same time. That gives teams a more accurate picture of what is wrong and what has changed.

Common error reasons

Error reason	Meaning
Overdue	The expected check-in did not arrive in time
Threshold Percent	Success rate in the current window dropped too low
Threshold Consecutive	Too many failures happened in a row
Expected Machines	Fewer than the required number of workers reported
Deactivate-By Deadline	The run did not deactivate before the finish deadline

Natural recovery vs manual reset

Natural recovery

Workflow behavior resolves the issue

A required check-in arrives, success rate improves, or another rule naturally returns to good standing.

Manual reset

Clear Errors acknowledges the issue

Use Clear Errors when the team wants to explicitly reset the condition without changing whether it is active or inactive.

Persistence matters

If a watch is deactivated while still in error, the issue remains visible. This is intentional. Deactivation stops monitoring, but it should not silently erase the fact that something went wrong.

Alerts and updates

Automation Watchdog can notify teams when:

• a watch enters error
• additional error reasons are added
• some error reasons clear while others remain
• the condition fully returns to relax

Practical guidance

• Use deactivate when monitoring for the current run should stop.
• Use Clear Errors when the team wants to acknowledge and reset unresolved state.
• Let the workflow recover naturally when you want the monitoring history to show what actually resolved the problem.

Read next

• Alert Behavior
• Terminology and Glossary