Choose rules deliberately

Owners and operators can edit monitoring; viewers can inspect it. Set shared defaults in Policies or open a machine's monitoring settings for an override. Offline, disk, CPU, memory and service rules are independent. Muting the offline bell does not mute the other enabled rules.

Offline notification timing

A device becomes Offline after the check-in grace period and evaluation. The incident begins at the actual transition; it is not backdated to the last check-in. Notification choices are Immediately, After 30 minutes, Immediately and after 30 minutes, and Muted.

Timing is captured when the incident opens. Later timing edits apply to the next incident, except muting cancels unsent offline notifications immediately. Unmuting does not recreate canceled messages. A fresh accepted check-in records recovery and cancels pending stages.

Low disk

Low disk uses two fresh samples below 10 GB or 5% free. Recovery requires two fresh samples above 15 GB and 7% free. These different trigger and clear conditions reduce repeated open/close cycles around one value. The low-disk email switch is separate from offline timing. Reporting covers the system disk only.

CPU and memory

Both resource alerts are disabled by default. Starting values are usage at or above 90% for five minutes, clearing at or below 80% for two minutes. Each resource has its own enabled state, high threshold/window and clear threshold/window. Durations accept whole minutes from 1 to 60; the clear threshold must be lower than the high threshold.

Accepted samples must be no more than 90 seconds apart. A missing metric, brief spike or longer gap resets a pending timing window. Faster check-ins cannot bypass elapsed time. Missing data never implies recovery. Values between the high and clear thresholds leave an existing incident open.

Changing an effective resource rule closes its incident as Rule changed and resets timing. Disabling it closes as Alert disabled. These are not recorded as successful recoveries. Saving identical effective values preserves timing.

Notifications and delivery evidence

Each resource or selected-service incident can submit one message. Recovery is recorded in the portal without a recovery email. Check Notifications for status and attempt history. An accepted SMTP submission is not evidence of delivery to a recipient's mailbox.

Only private development-inbox email has been exercised. Production customer email and delivery/bounce tracking remain pending. Maintenance pauses notification delivery while observations and incident history continue. A message already being submitted cannot be recalled.