Postgresql - Using the snapshots example from the docs with TestMain()

[purpose-robot]

Hello there,

I am trying to implement the Snapshots example from the docs using the TestMain() function.

My repository looks like this:

// repository.go
package repository

import "github.com/jackc/pgx/v5/pgxpool"

type Repository struct {
	Accounts AccountRepo
}

func New(pool *pgxpool.Pool) *Repository {
	return &Repository{
		Accounts: AccountRepo{Client: pool},
	}
}

For testing purposes, I created the following two files:

// test_main.go
package repository_test

import (
	"context"
	"log"
	"os"
	"path/filepath"
	"testing"
	"time"

	"github.com/jackc/pgx/v5/pgxpool"
	"github.com/testcontainers/testcontainers-go"
	"github.com/testcontainers/testcontainers-go/modules/postgres"
	"github.com/testcontainers/testcontainers-go/wait"

	"github.com/link_to_my_github_repository_goes_here/internal/repository"
)

var (
	ctx       = context.Background()
	repo      *repository.Repository
	container *postgres.PostgresContainer
)

func TestMain(m *testing.M) {
	var err error

	container, err = postgres.RunContainer(
		ctx,
		testcontainers.WithImage("postgres:16"),
		postgres.WithDatabase("test-db"),
		postgres.WithUsername("test-user"),
		postgres.WithPassword("test-password"),
		postgres.WithInitScripts(
			filepath.Join(".", "testdata", "init-db.sql"),
		),
		testcontainers.WithWaitStrategy(
			wait.ForLog("database system is ready to accept connections").
				WithOccurrence(2).
				WithStartupTimeout(5*time.Second),
		),
	)
	if err != nil {
		log.Fatalln("could not connect to Postgres database", err)
	}

	err = container.Snapshot(ctx, postgres.WithSnapshotName("test"))
	if err != nil {
		log.Fatalln("could not create Postgres database snapshot", err)
	}

	connString, err := container.ConnectionString(ctx, "sslmode=disable")
	if err != nil {
		log.Fatalln("could not retrieve Postgres connection string", err)
	}

	pool, err := pgxpool.New(ctx, connString)
	if err != nil {
		log.Fatalln("unable to create new Postgres connection pool", err)
	}
	defer pool.Close()

	if err := pool.Ping(ctx); err != nil {
		log.Fatalln("unable to aquire new Postgres connection pool", err)
	}

	repo = repository.New(pool)

	exitCode := m.Run()

	if err := container.Terminate(ctx); err != nil {
		log.Fatalln("tests completed, but failed to terminate container", err)
	}

	os.Exit(exitCode)
}

// accounts_test.go
package repository_test

import (
	"context"
	"testing"

	"github.com/google/uuid"
	"github.com/stretchr/testify/assert"
	"github.com/testcontainers/testcontainers-go/modules/postgres"

	"github.com/link_to_my_github_repository_goes_here/internal/repository"
)

func TestAccountInsert(t *testing.T) {
	tcs := []struct {
		name   string
		input  *repository.Account
		setup  func(ctx context.Context) error
		errors error
	}{
		{
			name: "Happy_Path",
			input: &repository.Account{
				Name:     "Morty Smith",
				Email:    "morty.smith@example.com",
				Password: []byte("013d7d16d7ad4fefb61bd95b765c8ceb"),
			},
			errors: nil,
		},
		{
			name: "With_Error_Duplicate_Email",
			input: &repository.Account{
				Name:     "Morty Smith",
				Email:    "morty.smith@example.com",
				Password: []byte("013d7d16d7ad4fefb61bd95b765c8ceb"),
			},
			setup: func(ctx context.Context) error {
				return repo.Accounts.Insert(&repository.Account{
					Name:     "Morty Smith",
					Email:    "morty.smith@example.com",
					Password: []byte("013d7d16d7ad4fefb61bd95b765c8ceb"),
				})
			},
			errors: repository.ErrDuplicateEmail,
		},
	}

	for _, tc := range tcs {
		t.Run(tc.name, func(t *testing.T) {
			if tc.setup != nil {
				assert.NoError(t, tc.setup(ctx))
			}

			err := repo.Accounts.Insert(tc.input)
			if tc.errors != nil {
				assert.ErrorIs(t, err, tc.errors)
			}

			if tc.errors == nil {
				assert.NoError(t, err)
				assert.IsType(t, uuid.UUID{}, tc.input.ID)
			}
		})

		t.Cleanup(func() {
			assert.NoError(t, container.Restore(ctx), postgres.WithSnapshotName("test"))
		})
	}
}

If I ran each test case separate, everything works as expected. However if I try to run both testcases one after another, it looks like the database snapshot is not working and the current table is still populated with data from my first test case. On the console, I got the following error messages:

2024/04/28 19:29:51 github.com/testcontainers/testcontainers-go - Connected to docker: 
  Server Version: 25.0.3
  API Version: 1.44
  Operating System: Debian GNU/Linux 12 (bookworm)
  Total Memory: 3921 MB
  Resolved Docker Host: unix:///run/user/1000/docker.sock
  Resolved Docker Socket Path: /run/user/1000/docker.sock
  Test SessionID: f6cb1512f303b7be781c0ee5855ae4c69a9a4acca8e28f0782dc49654f3476d3
  Test ProcessID: eb35c239-e65a-4914-9fb0-1ce4f5e62920
2024/04/28 19:29:51 🐳 Creating container for image testcontainers/ryuk:0.7.0
2024/04/28 19:29:51 ✅ Container created: 41e31db4955b
2024/04/28 19:29:51 🐳 Starting container: 41e31db4955b
2024/04/28 19:29:52 ✅ Container started: 41e31db4955b
2024/04/28 19:29:52 🚧 Waiting for container id 41e31db4955b image: testcontainers/ryuk:0.7.0. Waiting for: &{Port:8080/tcp timeout:<nil> PollInterval:100ms}
2024/04/28 19:29:52 🔔 Container is ready: 41e31db4955b
2024/04/28 19:29:52 🐳 Creating container for image postgres:16
2024/04/28 19:29:52 ✅ Container created: 32bb10138989
2024/04/28 19:29:52 🐳 Starting container: 32bb10138989
2024/04/28 19:29:52 ✅ Container started: 32bb10138989
2024/04/28 19:29:52 🚧 Waiting for container id 32bb10138989 image: postgres:16. Waiting for: &{timeout:<nil> deadline:0x400043eee8 Strategies:[0x40003c30e0]}
2024/04/28 19:29:53 🔔 Container is ready: 32bb10138989
=== RUN   TestAccountInsert
=== RUN   TestAccountInsert/Happy_Path
=== RUN   TestAccountInsert/With_Error_Duplicate_Email
    acounts_test.go:51: 
                Error Trace:    /home/hans/project_folder/internal/repository/acounts_test.go:51
                Error:          Received unexpected error:
                                duplicate email
                                ERROR: duplicate key value violates unique constraint "accounts_email_key" (SQLSTATE 23505)
                Test:           TestAccountInsert/With_Error_Duplicate_Email
--- FAIL: TestAccountInsert (1.31s)
    --- PASS: TestAccountInsert/Happy_Path (0.27s)
    --- FAIL: TestAccountInsert/With_Error_Duplicate_Email (0.53s)
FAIL
2024/04/28 19:29:54 🐳 Terminating container: 32bb10138989
2024/04/28 19:29:54 🚫 Container terminated: 32bb10138989
FAIL    github.com/link_to_my_github_repository_goes_here/internal/repository   3.124s
FAIL

My Go version is:

go version go1.22.0 linux/arm64

I am still a beginner with Go, so I am not 100 percent certain if this issue is related to testcontainers or if I just made a mistake when creating my testcases. Does someone have a working example which I could check out or could give me a hint about my problem?

[mdelapenya]

Hi @purpose-robot I think the error is in the cleanup function, which is executed in the outer t:

t.Cleanup(func() {
	assert.NoError(t, container.Restore(ctx), postgres.WithSnapshotName("test"))
})

Try moving it to the inner t, inside the for-loop here:

t.Run(tc.name, func(t *testing.T) 

Then the restore will execute for each subtest

[purpose-robot]

Hi @mdelapenya I changed the cleanup function as suggested, however now I got the following error message:

2024/04/30 12:10:04 github.com/testcontainers/testcontainers-go - Connected to docker: 
  Server Version: 25.0.3
  API Version: 1.44
  Operating System: Debian GNU/Linux 12 (bookworm)
  Total Memory: 3921 MB
  Resolved Docker Host: unix:///run/user/1000/docker.sock
  Resolved Docker Socket Path: /run/user/1000/docker.sock
  Test SessionID: 82074ef47ae09b7f34e2032a09f9676b677941eaf3052c6578409d5206b4a9ea
  Test ProcessID: a7c0bbd3-f329-48f4-9df2-e87b61ec0678
2024/04/30 12:10:04 🐳 Creating container for image testcontainers/ryuk:0.7.0
2024/04/30 12:10:04 ✅ Container created: d520f90f0470
2024/04/30 12:10:04 🐳 Starting container: d520f90f0470
2024/04/30 12:10:04 ✅ Container started: d520f90f0470
2024/04/30 12:10:04 🚧 Waiting for container id d520f90f0470 image: testcontainers/ryuk:0.7.0. Waiting for: &{Port:8080/tcp timeout:<nil> PollInterval:100ms}
2024/04/30 12:10:04 🔔 Container is ready: d520f90f0470
2024/04/30 12:10:04 🐳 Creating container for image postgres:16
2024/04/30 12:10:04 ✅ Container created: cd549bb707de
2024/04/30 12:10:04 🐳 Starting container: cd549bb707de
2024/04/30 12:10:04 ✅ Container started: cd549bb707de
2024/04/30 12:10:04 🚧 Waiting for container id cd549bb707de image: postgres:16. Waiting for: &{timeout:<nil> deadline:0x400043d228 Strategies:[0x40003c5200]}
2024/04/30 12:10:05 🔔 Container is ready: cd549bb707de
=== RUN   TestAccountInsert
=== RUN   TestAccountInsert/Happy_Path
=== RUN   TestAccountInsert/With_Error_Duplicate_Email
    acounts_test.go:51: 
                Error Trace:    /home/hans/project_folder/internal/repository/acounts_test.go:51
                Error:          Received unexpected error:
                                FATAL: terminating connection due to administrator command (SQLSTATE 57P01)
                Test:           TestAccountInsert/With_Error_Duplicate_Email
--- FAIL: TestAccountInsert (0.84s)
    --- PASS: TestAccountInsert/Happy_Path (0.57s)
    --- FAIL: TestAccountInsert/With_Error_Duplicate_Email (0.27s)
FAIL
2024/04/30 12:10:06 🐳 Terminating container: cd549bb707de
2024/04/30 12:10:06 🚫 Container terminated: cd549bb707de
FAIL    github.com/link_to_my_github_repository_goes_here/internal/repository   2.641s
FAIL

[mdelapenya]

Without the code it's very difficult to troubleshoot the error. By that message terminating connection due to administrator command (SQLSTATE 57P01) Have you looked up that error, which seems not related to testcontainers-go?

[purpose-robot]

I checked the error message and it means that the Postgres database was accidentally terminated.

As far as I can tell, this happens right after the call to container.Restore(ctx), so it might still be related to testcontainers-go.

I created a GitHub repository with a working example of my code which you can find here: Link to repo

By the way, I checked the example mentioned here and in my case the second test case fails too, so not sure what's wrong with me.

[purpose-robot]

Okay, I could solve my issue.
It was not related to testcontainers-go but to the way I initialise the database pool.

[mdelapenya]

Thanks for your time using the library! 🙇