Error when there are a lot of tests

[sazzle2611]

Hi

We use memongo in our GitLab CI pipeline to run tests before creating an image. Image is Alpine 3.21 with golang 1.24.

[memongo] [INFO]  Starting MongoDB with options &memongo.Options{ShouldUseReplica:true, Port:36343, CachePath:"", MongoVersion:"8.0.0", DownloadURL:"", MongodBin:"/usr/bin/mongod", Logger:(*log.Logger)(nil), LogLevel:0, StartupTimeout:10000000000, Auth:false}

--- FAIL: TestAddVehicle (31.78s)
    --- FAIL: TestAddVehicle/Error (30.90s)
        mongo_test.go:194: connection(localhost:36343[-85]) socket was unexpectedly closed: EOF
--- FAIL: TestUpdateVehicle (10.00s)
    --- FAIL: TestUpdateVehicle/OK (10.00s)
        --- FAIL: TestUpdateVehicle/OK/valid_data (10.00s)
panic: server selection error: context deadline exceeded, current topology: { Type: Unknown, Servers: [{ Addr: localhost:36343, Type: Unknown, Last error: dial tcp [::1]:36343: connect: connection refused }, ] } [recovered]
	panic: server selection error: context deadline exceeded, current topology: { Type: Unknown, Servers: [{ Addr: localhost:36343, Type: Unknown, Last error: dial tcp [::1]:36343: connect: connection refused }, ] }

When this error first appeared we used memongo v0.10.0 but I tried updating to v0.12.0 and there's same issue.

I commented out a lot of the tests and then it passed which is why I think it's related to that, I didn't comment out the tests that you see in the error message so it can't be related to them.

Any help on this would be greatly appreciated.

[saniales]

Hello, can you please provide more details in the kind of tests that you are trying to run? It appears to be an error related to MongoDB version 8.0. We are working on a fix.

[saniales]

Fixed by latest pr #43

[sazzle2611]

@saniales so sorry I missed the emails from this issue until I just came to check.

I just tried updating to v0.13.1 and re-enabling the tests I'd commented out but got the same error.

I don't think MongoDB version 8.0 is related as I have other services from the same project (meaning same sort of tests but different CI pipeline as we use microservices) that are using version 6.2.0 and they have the same issue, tried updating them as well to mongo 8.0 and memongo 0.13.1 but they still failed.

I'm not 100% sure what your asking for on the kind of tests we're running but they are basically API programs with FetchById, Find (search), Create, Update, Delete and occasionally a couple of other possible functions. Sometimes there can be quite a lot of tests as we try to test all possible scenarios for each function and things like adding Invoices have many ifs

Happy to provide you with more info if you let me know what you want.

Just realised that I didn't mention in first message that the tests run fine locally, this is odd as previously I had the problem where the tests wouldn't run locally (same error) so I commented them out to be able to check any new tests locally but I never pushed the commented out tests and the GitLab CI pipeline was fine running ALL the tests. So the roles have been reversed

[saniales]

Can you please share a reproducible test?

I need code and gitlab ci pipeline so i can run a local runner to test it myself. Thank you.

[sazzle2611]

@saniales I've created a reproducible test and attached a zip of the files, the tests run fine locally but not in CI pipeline. Locally I use Ubuntu 25.10

memongo-test.zip

[saniales]

Please share a test repository or a gist for security reasons. Thank you for your understanding.

[sazzle2611]

Sorry but I don't understand, what is the security issue? There is nothing private in those files.

We use our own self hosted GitLab and for security we don't allow logins from anyone outside our company.

[saniales]

Ok no problem we will check it out ASAP

[sazzle2611]

Thank you!!

[saniales]

I added a gist with the code of your zip so I can reference it

https://gist.github.com/saniales/946294a5cfdf298f6c4b2ad045a11ae0

My first question is about gitlab-ci.yml: What is the base image? I am assuming alpine.

Is there any particular reason why CGO is enabled?

What is the OS of the runner? which runner engine does it use? (docker, docker+machine, machine, other)

Please let me know so we can gather more intel on this :)

[sazzle2611]

Yes it's alpine, it's a golang image from dockerhub https://hub.docker.com/_/golang

To be honest I can't remember why we enabled CGO, was so long ago, but I know it was from some issue we had without it enabled. Once it worked we didn't touch it again

Host OS is RedHat and it's running in kubernetes

[saniales]

memongo/memongo_test.go

Lines 39 to 62 in b0447ce

	func TestWithReplica(t *testing.T) {
	versions := []string{"6.0.0", "7.0.0", "8.0.0"}
	for _, version := range versions {
	t.Run(version, func(t *testing.T) {
	server, err := memongo.StartWithOptions(&memongo.Options{
	MongoVersion: version,
	LogLevel: memongolog.LogLevelDebug,
	ShouldUseReplica: true,
	})
	require.NoError(t, err)
	defer server.Stop()
	uri := fmt.Sprintf("%s%s", server.URI(), "/retryWrites=false")
	client, err := mongo.Connect(context.Background(), options.Client().ApplyURI(uri))
	if err != nil {
	t.Logf("err Connect: %v", err)
	}
	require.NoError(t, err)
	require.NoError(t, client.Ping(context.Background(), readpref.Primary()))
	})
	}
	}

Check the usage in this test, can you try to add "/retryWrites=false" at the end of the URI? just to be sure