Question - Best Practice of Test "service"

[bt-Knodel]

The documentation for datadog ci says the following about setting service for tests:

service
Name of the service or library under test.
Environment variable: DD_SERVICE
Default: (test framework name)
Example: my-ui

Knowing that, what is the best practice for setting service? For example, if we host a site of "somesite.com", should tests against "somesite.com" be labeled with the service "somesite.com"? That's what I interpret the verbiage of "Name of the service or library under test." to be. It does further explain in the docs that "session" should be used to differentiate between tests against the same service, but one of the problems there is that flaky test management doesn't allow filtering on session.

Apologies if an issue was the incorrect way to get this clarification, but needed to ask.

[peoplenarthax]

Hello @bt-Knodel , so Test Service is a label that will help in use cases like monorepos. We recommend setting different Test Service names for different logical groupings of tests within the same monorepo.

So for instance in a monorepo where you divide ownership, or you want to declare boundaries across the services there, you could add the tag of test service for the test of that specific area.

This tag is optional, so you do not need to specify it if you see no use of it.

[bt-Knodel]

@peoplenarthax , sorry on the delay.

So, our structure for example is all our tests are around a single site (Ex. "somesite.com"). We have a couple test suites:

• End to end
• Service Tests
• Unit Tests

I read the documentation as all of these should be the same service "somesite.com", but separated by the intention. So, under service "somesite.com":

• session: e2e
• session: unit
• session: integration

Is that the most appropriate, or should each test suite really be labeled as a distinct service rather than all under "somesite.com"?