Service Monitoring

Entry: /monitoring. Monitoring owns service probes, device observations, alert rules, maintenance and incidents. You need the appropriate Group and monitor or collection-management permission. Confirm that the installation can reach the approved target and that required workers, adapters and credentials are configured; saving a target alone does not prove collection is operating.

Create a service monitor#

1. Open Monitors and choose Create Monitor. Enter a name and select the supported probe kind, such as HTTPS, TCP, DNS or TLS.
2. Enter the target and set the interval, timeout and applicable expected status or content assertion. Service-monitor intervals range from 60 seconds to 86,400 seconds; the timeout must satisfy the form's validation.
3. Submit, open the monitor detail and inspect Recent checks, status and observation time after the worker has run.
4. Use Availability or the monitor's Availability report to review the chosen reporting window. Missing or failed collection is not a successful check.

Configure alerting and handle an incident#

1. In the monitor's permitted operations, configure its alert rule: failures before an incident, successes before automatic resolution, severity and any escalation policy.
2. Schedule Maintenance for planned work with the correct start, end and reason. Confirm the saved window before relying on its alerting effect.
3. Inspect Incident history when an incident occurs. Use Acknowledge Incident when you take responsibility, and Resolve Incident with a meaningful note after verification.
4. Check the actual configured notification destination. Selecting an email or webhook channel does not itself configure an external delivery provider.

Configure device performance collection#

If the device needs vendor-specific readings, an authorized operator first opens Vendor MIBs, selects the enabled Device Model from the Group's Asset catalog, chooses its base profile and defines the numeric OIDs. Uploading a MIB file is reference evidence only; it does not create mappings automatically.

1. Open Collection Settings → Add collection target. Choose the standalone or eligible inventory source, target details, Device profile and required collection settings, and select the Device Model so the target inherits its mappings.
2. Save the target, then use its Settings to select the supported CPU, memory, interface, power-supply and fan components. Review available named interfaces rather than assuming every component is supported.
3. Choose Save collection settings, request Check now when available, and inspect Network Performance and its component history. Use Device Logs and Traffic Analysis for configured receiver evidence.

Limits and boundaries: device customization is limited to 512 selected components per target; 32 custom definitions per model and per effective target; a 256 KiB customization bound per target and per model profile; manual collection has a 60-second cooldown. Vendor/firmware coverage varies. Custom MIB definitions require verified device knowledge; ICMP requires its separate configured adapter. Managed/Unmanaged inventory state can stop linked collection. Monitoring observes targets; it does not provision them, change their inventory or guarantee physical-device compatibility from a generic profile.

Schedule on-call and publish a status page#

1. Create an on-call schedule with a time zone, up to five layers of 1 to 50 participants and up to 200 overrides. Participants and overrides must be active Users in the Group. The schedule shows who is on call and the next 14 days of hand-offs.
2. In an alert rule's escalation policy, target a schedule so the notification goes to the User on call at that time. A stage whose schedule has been retired is skipped with a timeline note, and the next stage proceeds on its own delay.
3. To show service state to people without an account, publish up to 50 Monitors under display names at /status/<slug> on your installation. If a slug is taken or reserved you see the same generic message, so other Groups' slugs cannot be discovered.
4. Open the page without signing in. It shows only state and 90-day daily uptime. Unpublishing removes it immediately. Public requests are limited to 60 per minute per source address.

For a service that should be checked from several places, open Monitor Settings and choose up to 10 probe locations and a failure quorum. A round is Down only when the quorum of locations fails, so detection can lag by up to one interval.