TCP Check

Agent CheckAgent Check
Network Graphs


Monitor TCP connectivity and response time for any host and port.



The TCP check is packaged with the Agent, so simply install the Agent on any host from which you want to probe TCP ports. Though many metrics-oriented checks are best run on the same host(s) as the monitored service, you’ll probably want to run this check from hosts that do not run the monitored TCP services, i.e. to test remote connectivity.

If you need the newest version of the TCP check, install the dd-check-tcp package.


Create a file tcp_check.yaml in the Agent’s conf.d directory. See the sample tcp_check.yaml for all available configuration options:


  - name: SSH check
    host: # or an IPv4/IPv6 address
    port: 22           
    skip_event: true # if false, the Agent will emit both events and service checks for this port; recommended true (i.e. only submit service checks)
    collect_response_time: true # to collect network.tcp.response_time. Default is false.

Configuration Options

  • name (Required) - Name of the service. This will be included as a tag: instance:<name>. Note: This tag will have any spaces and dashes converted to underscores.
  • host (Required) - Host to be checked. This will be included as a tag: url:<host>:<port>.
  • port (Required) - Port to be checked. This will be included as a tag: url:<host>:<port>.
  • timeout (Optional) - Timeout for the check. Defaults to 10 seconds.
  • collect_response_time (Optional) - Defaults to false. If this is not set to true, no response time metric will be collected. If it is set to true, the metric returned is network.tcp.response_time.
  • skip_event (Optional) - Defaults to false. Set to true to skip creating an event. This option will be removed in a future version and will default to true.
  • tags (Optional) - Tags to be assigned to the metric.

Restart the Agent to start sending TCP service checks and response times to Datadog.


Run the Agent’s info subcommand and look for tcp_check under the Checks section:


      - instance #0 [OK]
      - Collected 1 metric, 0 events & 1 service check



The TCP check is compatible with all major platforms.

Data Collected


The response time of a given host and TCP port, tagged with url, e.g. 'url:'.
shown as second


The TCP check does not include any event at this time.

Service Checks


Returns DOWN if the Agent cannot connect to the configured host and port, otherwise UP.

Older versions of the TCP check only emitted events to reflect changes in connectivity. This was eventually deprecated in favor of service checks, but you can still have the check emit events by setting skip_event: false.

To create alert conditions on this service check in the Datadog app, click Network on the Create Monitor page, not Integration.


Need help? Contact Datadog Support.

Further Reading

Learn more about infrastructure monitoring and all our integrations on our blog