Grafana Cloud

Check types

Synthetic Monitoring checks are tests that run on selected probes at defined intervals and report metrics and logs back to your Grafana Cloud account. The target for checks can be a domain name, a server, or a website, depending on what information you would like to gather about your endpoint. You can define multiple checks for a single endpoint to check different capabilities.

All checks have a “job name” and “target”, regardless of type. This pair of values is used to identify each check uniquely. In the resulting metrics, these correspond to the value of the job and instance labels, respectively. The “job name” can be used to identify the resources associated with the check (for example, “production servers”) or the goal of the validation (for example, “POST is not accepted”). Keep in mind that some check types might use the same value for their target (for example, ping and DNS checks require the target to be a hostname), and the “job name” is the only way to make them unique.

Supported checks

Below is some basic information about each type of check supported by Synthetic Monitoring. When creating checks, click Options to see advanced settings for each type.


Ping is the simplest check to test that an endpoint is available. Target servers must be reachable from the probe’s network and be configured to reply to ICMP echo requests. The endpoint’s response time is used to measure latency from each probe location.


HTTP and HTTPS checks are used to test websites. Uptime and response latency are measured like other check types. These checks can also be configured for more advanced tests, like if a site is using a specific version of SSL, for SSL certificate expiration, or if HTTP automatically redirects to HTTPS.

You can also use the MultiHTTP check type to test multiple URLs in a single check.


DNS or domain name servers resolve a domain such as to an IP address. This check ensures a domain resolves and measures the average time for the resolution to happen. DNS checks can be set up to validate against a specific DNS server and check for a specific kind of response.


A TCP check connects to an endpoint on a given hostname or IP address and a port. This ensures a server accepts the connection and measures response latency and uptime.


A Traceroute check runs a traceroute from probes to targets. You can trace the path of a request through the internet and get more detailed information about how your requests are flowing through the public internet.

Traceroute checks can be useful to visualize a network path, see how that path is changing over time, and how it’s reaching the destination.

For more details, refer to How traceroute in the Synthetic Monitoring plugin for Grafana Cloud helps network troubleshooting.

Common options

The following is a list of options common to all check types. All these options are required:

Option nameDescriptionMetric label
EnabledWhether the check is enabled or not.none
Job nameThe name for the check.job
TargetThe target for the check, the value depends on check type.instance
Probe locationsThe locations where the check should run from.probe
FrequencyThe frequency the check should run in seconds. The value can range from 60 to 3600 seconds.frequency٭
TimeoutMaximum execution time for the check. The value can range from 1 to 60 seconds.none
Publish full set of metricsWhether to publish additional metrics for the check.none

The named labels, with the exception of frequency, are applied to all metrics. The frequency label is only applied to the sm_check_info metric.

Optionally, checks can include custom labels. You can specify both the name and the value of the labels. In the metrics, the name that you provided is prefixed with the string label_.

Check Metrics

Refer to Check Metrics for more details.

The “publish full set of metrics” option will cause additional metrics to be published for the check. Currently, these metrics are associated with the timings and can be used to create histograms.

Next steps

Refer to Set up private probes to test servers not reachable from public networks or test from different locations than the ones available.