Custom check monitors include any service check not reported by one of the more than 450 integrations included with the Agent. Custom service checks can be sent to Datadog using a custom Agent check, DogStatsD, or the API.
To create a custom check monitor in Datadog, use the main navigation: Monitors –> New Monitor –> Custom Check.
Choose a custom check from the drop-down box.
Select the scope to monitor by choosing host names, tags, or choose
All Monitored Hosts. If you need to exclude certain hosts, use the second field to list names or tags.
ANDlogic. All listed hostnames and tags must be present on a host for it to be included.
ORlogic. Any host with a listed hostname or tag is excluded.
In this section, choose between a Check Alert or Cluster Alert:
A check alert tracks consecutive statuses submitted per check grouping and compares it to your thresholds.
Set up the check alert:
Trigger a separate alert for each
<GROUP> reporting your check.
Check grouping is specified either from a list of known groupings or by you. For custom check monitors, the per-check grouping is unknown, so you must specify it.
Trigger the alert after selected consecutive failures:
Each check run submits a single status of
UNKNOWN. Choose how many consecutive runs with the
CRITICAL status trigger a notification. For example, to be notified immediately when your check fails, trigger the monitor alert on
1 critical status or
1 warning status.
Do not notify or
Notify for Unknown status.
Resolve the alert after selected consecutive successes:
Choose how many consecutive runs with the
OK status resolve the alert. For example, to ensure an issue is fixed, resolve the monitor on
4 OK statuses.
A cluster alert calculates the percent of checks in a given status and compares it to your thresholds.
Set up a cluster alert:
Decide whether or not to group your checks according to a tag.
Ungrouped calculates the status percentage across all sources.
Grouped calculates the status percentage on a per-group basis.
Select the percentage for alert and warn thresholds. Only one setting (alert or warn) is required.
For detailed instructions on the Say what’s happening and Notify your team sections, see the Notifications page.
Additional helpful documentation, links, and articles: