Skip running tests from Nebula Logger Unlocked package

[gauravchaube21]

We have installed Nebula Logger as an unlocked package in our org. Our pipeline installs the Nebula Logger to a scratch org whenever it gets created. Along with the package, we install the regular metadata and run test classes. This happens via our CI/CD Pipeline. We use the following coming to run the test classes
sf apex run test --test-level RunLocalTests --wait 30 --code-coverage --detailed-coverage --result-format human

Unfortunately, even though we have specified RunLocalTests in our sf command, all of the test classes run when running the command. Infact, a lot of test classes fail (Error Message: Search Unavailable) when this happens.

It would be great if someone could help us with the reason why the package test classes run even though we explicitly run local test classes and a possible workaround in this situation. Our pipelines are blocked because of the failing test classes within the package.

Thanks in advance

[jongpie]

Hi @gauravchaube21 - running Nebula Logger's tests is unfortunately expected behavior when using the unlocked package and --test-level RunLocalTests. In this case, the term 'local' refers to any Apex classes without a namespace - and Nebula Logger's unlocked package does not have a namespace, so it's considered 'local code', even though it comes from an unlocked package.

There was a similar question about this last year (see my comment here) - one option is for you to create a test suite containing your tests (and excluding Nebula Logger's tests). It's not really ideal, you'd have to maintain a test suite, but it's one way to avoid running Nebula Logger's tests.

As far as that specific error about 'Search unavailable' - I've been running into that a lot myself over the last few weeks, I think it might be related to some Salesforce infrastructure issues. I'm going to see if there's anything I can do in that test to avoid this issue - this single test method isn't that important, it shouldn't cause pipeline failures.

[gauravchaube21]

Thank you so much for your response, @jongpie. I really appreciate your suggestions for the issue. I'll try setting up a test suite for the time being. Meanwhile, do you think you would be able to pinpoint the test class problem and patch that out in the package?

[jongpie]

@gauravchaube21 I'm still exploring options for how to solve this - it's throwing an UnexpectedException, which makes this difficult to handle. I don't think these exceptions can be caught in try-catch block (Apex will still throw the exception), and I'd prefer not to eliminate the test method (because it provides valuable automated testing.... when it works 🙈 ).

So, I'm not sure what good options there are for handling this, but I'll try to explore some more options ASAP. If I can find a good approach, I'll release a new version with a fix.

[jongpie]

@gauravchaube21 I've created issue #624 to track this problem.

[jongpie]

@gauravchaube21 this has been "fixed" in release v4.13.0 - I decided to just comment-out the problematic test method 😅

I'll revisit this later, but the functionality being tested has not changed in a few years, so commenting-out the test method seems safe enough for now. Let me know if you run into any other issues!

[gauravchaube21]

Thank you so much @jongpie . I really appreciate your help on this.❤️

[jongpie]

No problem!