Overview

gRPC tests allow you to proactively monitor your gRPC services and servers. You can choose from two types:

Behavior Checks

Send gRPC requests to your applications’ API endpoints to verify responses and defined conditions, such as overall response time, header, or body content.

Health Checks

gRPC health checks are a standard for reporting the health of gRPC services. Determine if your gRPC servers and services are responsive, running, and capable of handling remote procedure calls (RPCs).

By implementing gRPC health checks, you can run gRPC health checks tests without having to provide a .proto file to Datadog. For more information, see the example health checks .proto file shared by the gRPC community.

gRPC tests can run from both managed and private locations depending on your preference for running the test from outside or inside your network. gRPC tests can run on a schedule, on-demand, or directly within your CI/CD pipelines.

Configuration

After choosing to create a gRPC test, define your test’s request.

Define request

1. Specify the Host and Port to run your test on. The default gRPC port is 50051.

2. Select Behavior Check to perform a unary call or Health Check to perform a health check.

   For a behavior check, specify the Server Reflection or upload a Proto File that defines your gRPC server. Select a method and include a request message. Datadog does not support streaming methods.

   For a health check, enter the name of the service. Leave this field blank if you want to send a health check on the gRPC server.

3. Add Advanced Options (optional) to your test:

   • Timeout: Specify the amount of time in seconds before the test times out.
   • Ignore server certificate error: Select to have your gRPC test go on with connection even if there are errors when validating the SSL certificate.
   • gRPC metadata: Add and define metadata to your gRPC request to pass metadata between services.

   • Client certificate: Authenticate through mTLS by uploading your client certificate (.crt) and the associated private key (.key) in PEM format.

     
     

     You can use the openssl library to convert your certificates. For example, convert a PKCS12 certificate to PEM formatted private keys and certificates.

     openssl pkcs12 -in <CERT>.p12 -out <CERT_KEY>.key -nodes -nocerts
     openssl pkcs12 -in <CERT>.p12 -out <CERT>.cert -nokeys
     

4. Name your gRPC test.

5. Add env Tags as well as any other tag to your gRPC test. You can then use these tags to filter through your Synthetic tests on the Synthetic Monitoring & Continuous Testing page.

Click Send to try out the request configuration. A response preview is displayed on the right side of your screen.

Snippets

When setting up a new Synthetic Monitoring API test, use snippets to automatically fill in basic auth, performance, and regions, rather than selecting these options manually. The following snippets are available:

• Basic Auth: Automatically test your APIs using pre-populated basic auth headers, JavaScript, bearer token, and API/app key auth variables.

• Performance: Automatically configure a test with the shortest frequency (one minute), perform a gRPC health check, and test for overall response time latency with a breakdown of network timing.

• Regions: Automatically test your API endpoint against a location in each of the three primary geographic regions (AMER, APAC and EMEA).
  
  

  

Define assertions

Assertions define what an expected test result is. After you click Send, an assertion on the response time is added based on the response that was obtained. You must define at least one assertion for your test to monitor.

Type	Operator	Value type
response time	is less than	Integer (ms)
gRPC response	contains, does not contain, is, is not, matches, does not match, jsonpath, xpath	String Regex
gRPC metadata	is, is not, contains, does not contain, matches regex, does not match regex, does not exist	Integer (ms) Regex

You can create up to 20 assertions per API test by clicking New Assertion or by clicking directly on the response preview:

Type	Operator	Value type
response time	is less than	Integer (ms)
healthcheck status	is, is not	Integer (ms)
gRPC metadata	is, is not, contains, does not contain, matches regex, does not match regex, does not exist	Integer (ms)

You can create up to 20 assertions per API test by clicking New Assertion or by clicking directly on the response preview:

If a test does not contain an assertion on the response body, the body payload drops and returns an associated response time for the request within the timeout limit set by the Synthetics Worker.

If a test contains an assertion on the response body and the timeout limit is reached, an Assertions on the body/response cannot be run beyond this limit error appears.

Select locations

Select the Locations to run your gRPC test from. gRPC tests can run from both managed and private locations depending on your preference for running the test from outside or inside your network.

Datadog’s out-of-the-box managed locations allow you to test public-facing websites and endpoints from regions where your customers are located.

The Datadog for Government site (US1-FED) uses the following managed location:

Specify test frequency

gRPC tests can run:

• On a schedule to ensure your most important services are always accessible to your users. Select the frequency at which you want Datadog to run your gRPC test.
• Within your CI/CD pipelines to start shipping without fearing faulty code might impact your customers experience.
• On-demand to run your tests whenever makes the most sense for your team.

Define alert conditions

Set alert conditions to determine the circumstances under which you want a test to fail and trigger an alert.

Alerting rule

When you set the alert conditions to: An alert is triggered if any assertion fails for X minutes from any n of N locations, an alert is triggered only if these two conditions are true:

• At least one location was in failure (at least one assertion failed) during the last X minutes;
• At one moment during the last X minutes, at least n locations were in failure.

Fast retry

Your test can trigger retries X times after Y ms in case of a failed test result. Customize the retry interval to suit your alerting sensibility.

Location uptime is computed on a per-evaluation basis (whether the last test result before evaluation was up or down). The total uptime is computed based on the configured alert conditions. Notifications sent are based on the total uptime.

Configure the test monitor

A notification is sent by your test based on the alerting conditions previously defined. Use this section to define how and what to message your team.

1. Similar to how you configure monitors, select users and/or services that should receive notifications either by adding an @notification to the message or by searching for team members and connected integrations with the dropdown menu.

2. Enter the notification message for your test. This field allows standard Markdown formatting and supports the following conditional variables:

   Conditional Variable	Description
   {{ #is_alert }}	Show when the test alerts.
   {{ ^is_alert }}	Show unless the test alerts.
   {{ #is_recovery }}	Show when the test recovers from alert.
   {{ ^is_recovery }}	Show unless the test recovers from alert.
   {{ #is_renotify }}	Show when the monitor renotifies.
   {{ ^is_renotify }}	Show unless the monitor renotifies.
   {{ #is_priority }}	Show when the monitor matches priority (P1 to P5).
   {{ ^is_priority }}	Show unless the monitor matches priority (P1 to P5).

3. Specify how often you want your test to re-send the notification message in case of test failure. To prevent renotification on failing tests, leave the option as Never renotify if the monitor has not been resolved.

4. Click Create to save your test configuration and monitor.

For more information, see Using Synthetic Test Monitors.

Create local variables

To create a local variable, click Create a Local Variable. You can select one of the following available builtins to add to your variable string:

{{ numeric(n) }}

Generates a numeric string with n digits.

{{ alphabetic(n) }}

Generates an alphabetic string with n letters.

{{ alphanumeric(n) }}

Generates an alphanumeric string with n characters.

{{ date(n unit, format) }}

Generates a date in one of Datadog’s accepted formats with a value corresponding to the UTC date the test is initiated at + or - n units.

{{ timestamp(n, unit) }}

Generates a timestamp in one of Datadog’s accepted units with a value corresponding to the UTC timestamp the test is initiated at +/- n units.

{{ uuid }}

Generates a version 4 universally unique identifier (UUID).

{{ public-id }}

Injects the Public ID of your test.

{{ result-id }}

Injects the Result ID of your test run.

To obfuscate local variable values in test results, select Hide and obfuscate variable value. Once you have defined the variable string, click Add Variable.

Use variables

You can use the global variables defined on the Settings page in the URL, advanced options, and assertions of your gRPC tests.

To display your list of variables, type {{ in your desired field.

Test failure

A test is considered FAILED if it does not satisfy one or more assertions or if the request prematurely failed. In some cases, the test can fail without testing the assertions against the endpoint.

These reasons include the following:

gRPC specific errors

gRPC has a list of specific status codes that can be found in the official gRPC documentation.

CONNRESET

The connection was abruptly closed by the remote server. Possible causes include the web server encountering an error or crashing while responding, or losing connectivity to the web server.

DNS

DNS entry not found for the test URL. Possible causes include a misconfigured test URL or the wrong configuration of your DNS entries.

INVALID_REQUEST

The configuration of the test is invalid (for example, a typo in the URL).

SSL

The SSL connection couldn’t be performed. See the dedicated error page for more information.

TIMEOUT

The request couldn’t be completed in a reasonable time. Two types of TIMEOUT can happen:

• TIMEOUT: The request couldn't be completed in a reasonable time. indicates that the request duration hit the test defined timeout (default is set to 60 seconds). For each request, only the completed stages for the request are displayed in the network waterfall. For example, in the case of Total response time only being displayed, the timeout occurred during the DNS resolution.
• TIMEOUT: Overall test execution couldn't be completed in a reasonable time. indicates that the test duration (request and assertions) hits the maximum duration of 60.5 seconds.

Permissions

By default, only users with the Datadog Admin and Datadog Standard roles can create, edit, and delete Synthetic gRPC tests. To get create, edit, and delete access to Synthetic gRPC tests, upgrade your user to one of those two default roles.

If you are using the custom role feature, add your user to any custom role that includes synthetics_read and synthetics_write permissions.

Restrict access

Access restriction is available for customers using custom roles on their accounts.

You can restrict access to a browser test based on the roles in your organization. When creating a browser test, choose which roles (in addition to your user) can read and write your test.

Further Reading

Additional helpful documentation, links, and articles:

Introducing Datadog Synthetic MonitoringBLOG Monitor your gRPC APIs with DatadogBLOG Introduction to Synthetic TestsLEARNING CENTER Chain gRPC requests with multistep API testsDOCUMENTATION Run gRPC tests on internal endpointsDOCUMENTATION Learn about Synthetic test monitorsDOCUMENTATION