Select what should stay running
Choose up to ten exact names per platform. Share the selection through an organization or site policy, or replace it with a machine override. Service monitoring is off until you select names; it is not an automatic inventory of every installed service.
The reviewed capability requires Linux 0.9.0 or Windows 0.11.0. macOS services are not included.
Observe a sustained stop
Choose a stopped duration from one to sixty minutes. Fresh Stopped, Failed, Paused or Not installed observations can open separate incidents for each selected service. Unknown, stale, unsupported or transition observations do not fabricate failures or recovery.
A confirmed Running observation closes the incident. Polling can miss very brief failures between check-ins. Read the full timing behavior.
Preserve the boundary
Collectors query service state using restricted access. They do not restart services, execute scripts or automatically remediate an incident. Services that legitimately stop when idle need a different expectation; do not select them simply because they exist.
Begin with a meaningful test
The implementation has recorded disposable Linux and Windows service lifecycle tests. For your own evaluation, use an authorized disposable fixture, then check the observed state, incident, notification history and recovery.
Read the service-stop alert guide or configure your first selected service.