Files
docker-docs/content/guides/testcontainers-java-replace-h2/problem-with-h2.md
Manuel de la Peña b951e92f57 feat(guides): migrate all testcontainers.com guides (#24505)
## Description

Migrate 17 Testcontainers guides from testcontainers.com into the Docker
docs site, covering Java (14 guides), .NET (2 guides), and Node.js (1
guide). This follows up on PR #24450 which added the initial Go and
Python guides.

Each guide is converted from AsciiDoc to Hugo Markdown, split into
multi-chapter stepper navigation, updated to the latest Testcontainers
API, and verified with passing tests running in containers.

Java guides use testcontainers-java 2.0.4 with the new 2.x Maven
coordinates and package names (e.g., `testcontainers-postgresql`,
`org.testcontainers.postgresql.PostgreSQLContainer`). The Quarkus guide
uses Quarkus 3.22.3 with TC 1.x managed by the Quarkus BOM, since no
released Quarkus version ships TC 2.x yet.

## How to test

All code snippets have been verified by running each guide's source
repository tests inside Docker containers with the Docker socket
mounted.

To re-run the verification, use the `/testcontainers-guides-migrator`
skill included in this PR
(`.claude/skills/testcontainers-guides-migrator/SKILL.md`). The skill's
Step 6 documents the exact container commands and macOS Docker Desktop
workarounds (host override, docker-java API version, etc.) needed to run
each language's tests:

```
/testcontainers-guides-migrator I want you to verify all the guides in this branch.
Do a full review, verifying that all code snippets compile, the code is executable,
and ALL the tests pass. Run them as docker containers, never locally.
```

## Related issues or tickets

Supersedes #24450 (expanded from 2 guides to all 19)

## Reviews

- [ ] Technical review
- [ ] Editorial review
- [ ] Product review

---------

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-25 10:03:26 +00:00

62 lines
1.9 KiB
Markdown

---
title: The problem with H2 for testing
linkTitle: The H2 problem
description: Understand why using H2 in-memory databases for testing gives false confidence.
weight: 10
---
A common practice is to use lightweight databases like H2 or HSQL as
in-memory databases for testing while using PostgreSQL, MySQL, or Oracle in
production. This approach has significant drawbacks:
- The test database might not support all features of your production database.
- SQL syntax might not be compatible between H2 and your production database.
- Tests passing with H2 don't guarantee they'll work in production.
## Example: PostgreSQL-specific syntax
Consider implementing an "upsert" — insert a product only if it doesn't
already exist. In PostgreSQL, you can use:
```sql
INSERT INTO products(id, code, name) VALUES(?,?,?) ON CONFLICT DO NOTHING;
```
This query doesn't work with H2 by default:
```text
Caused by: org.h2.jdbc.JdbcSQLException: Syntax error in SQL statement
"INSERT INTO products (id, code, name) VALUES (?, ?, ?) ON[*] CONFLICT DO NOTHING";
```
You can run H2 in PostgreSQL compatibility mode, but not all features are
supported. The inverse is also true — H2 supports `ROWNUM()` which PostgreSQL
doesn't.
Testing with a different database than production means you can't trust your
test results and must verify after deployment, defeating the purpose of
automated tests.
## The Spring Boot test using H2
A typical H2-based test looks like this:
```java
@DataJpaTest
class ProductRepositoryTest {
@Autowired
ProductRepository productRepository;
@Test
@Sql("classpath:/sql/seed-data.sql")
void shouldGetAllProducts() {
List<Product> products = productRepository.findAll();
assertEquals(2, products.size());
}
}
```
Spring Boot uses H2 automatically when it's on the classpath. The test passes,
but it doesn't catch PostgreSQL-specific issues.