Quarkus H2 in-mem DB - how to connect while running in quarkus:dev mode

Hi Kawu, thank you for your question!

There is a public issue about this problem with connection to in-memory DB: https://youtrack.jetbrains.com/issue/DBE-23149
You are welcome to subscribe to the issue to get updates on its progress. See this article https://intellij-support.jetbrains.com/hc/en-us/articles/207241135 if you are not familiar with our bug tracking system.

As a temporary workaround, please use a local database or a remote database instead of an in-memory DB and add it as a data source in the Database tool window.
Thanks!

I was able to hook a TCP server around the H2 in-mem DB on port 9092 programmatically like:

```

import java.sql.SQLException;

import jakarta.enterprise.context.ApplicationScoped;
import jakarta.enterprise.event.Observes;

import io.quarkus.runtime.ShutdownEvent;
import io.quarkus.runtime.StartupEvent;

import org.h2.tools.Server;

@ApplicationScoped
public class ApplicationLifecycle {

    Server server;

    void onStart(@Observes StartupEvent event) throws SQLException {
        server = Server.createTcpServer("-tcpPort", "9092", "-tcpAllowOthers").start();
    }

    void onStop(@Observes ShutdownEvent event) {
        server.stop();
    }
}

```

This code was recklessly stolen from https://stackoverflow.com/questions/57802845/quarkus-jpa-with-h2-database

Is this possible in plain Quarkus (properties files)?

Kawu, Thank you for your update.
What you’re doing here is starting an H2 TCP server (org.h2.tools.Server) inside the Quarkus process during the application lifecycle. That’s H2-specific runtime logic, not a Quarkus feature that can be enabled purely through Quarkus datasource properties.

But you can use Quarkus Dev Services for databases (for example, with PostgreSQL because H2 is not supported as a container in Dev Services). In this case, you will be able to fully configure the database without using an additional class with logic as in your example.

To use this scenario, you need:

1. You must have it installed and running Docker in your machine with IntelliJ IDEA;

2. Add the following dependency in your pom.xml (or similar if you use Gralde):

   <dependency>
     <groupId>io.quarkus</groupId>
     <artifactId>quarkus-jdbc-postgresql</artifactId>
   </dependency>

3. In application.properties file specify database configuration e.g.:

   %dev.quarkus.datasource.db-kind=postgresql
   %dev.quarkus.datasource.devservices.port=5555
   %dev.quarkus.datasource.devservices.db-name=devdb
   %dev.quarkus.datasource.devservices.username=quarkus
   %dev.quarkus.datasource.devservices.password=quarkus
   %dev.quarkus.datasource.devservices.init-script-path=import.sql

4. In the dev profile you configure db-kind and devservices.* settings, and you should NOT set
   %dev.quarkus.datasource.jdbc.url , %dev.quarkus.datasource.username and etc. otherwise Dev Services will be disabled for that profile because Quarkus treats the datasource as explicitly configured.
5. In the Database tool window, add a new PostgreSQL Data Source using connection data from application.properties.

Please let me know if you have any questions or difficulties.

Thanks for the tip/s.

The “You must have Docker installed” part was a bigger extra step than expected, which had to be taken (with some problems). I really had to make BIOS changes to activate CPU virtualization, only then Docker could actually be launched.

Dev services seem to be running now, however I still have trouble with duplicate tables. I want to run a PostgreSQL SQL (DDL) script to create the tables instead of letting Hibernate do so from the entities. The entities do not contain any annotations to reflect the `String` (VARCHAR) length of the column nor the correct types etc. for other columns.

I thought to suppress Hibernate from creating the tables via `%dev.quarkus.hibernate-orm.schema-management.strategy = none`, but sth. is not working.

Let me check…

Thanks for your update.

Since Quarkus 3.23 you should use the quarkus.hibernate-orm.schema-management.strategy instead of quarkus.hibernate-orm.database.generation (https://quarkus.io/guides/hibernate-orm)
So in your case you can use: 
%dev.quarkus.hibernate-orm.schema-management.strategy=none
%test.quarkus.hibernate-orm.schema-management.strategy=none

And make sure that you have no properties like: %dev.quarkus.hibernate-orm.database.generation=drop-and-create

Use %dev.quarkus.datasource.devservices.init-script-path=db/init-dev.sql property to run db init script on Dev services startup.

If you still experience difficulties, please provide more us screenshot of the issue (e.g. Database tool window where we can see duplicate tables) and your application.properties file.
Upload them securely to our server https://uploads.jetbrains.com/ and provide us with the Upload ID.

The property:

%dev.quarkus.hibernate-orm.schema-management.strategy=none

seems to do the job now. Thanks so much!

PostgreSQL Docker is up an running in dev mode/profile with the following init scripts being picked up (I created a subdir “db” in the src/main/resources dir):

%dev.quarkus.datasource.devservices.init-script-path=db/create-tables-postgresql.sql, db/insert-data-postgresql.sql

I can see the correct data from the DS connection, manipulate data etc.

Two problems remain however:

1. The schema name is always “public”

From the create-tables-postgresql.sql:

CREATE TABLE IF NOT EXISTS "Countries" (
…
);

Prefixing the table names with a schema name like:

CREATE TABLE IF NOT EXISTS retract."Countries" (
…
);

leads to

Caused by: org.postgresql.util.PSQLException: ERROR: schema "retract" does not exist
 Position: 28

How do you launch a dev PG container giving “it” a specific schema name other than “public”?

See IJ connection:

2. At startup, Hibernate complains with “Failed to validate Schema”:

2026-03-11 13:54:36,340 ERROR [io.quarkus.hibernate.orm.runtime.schema.SchemaManagementIntegrator] (Hibernate post-boot validation thread for <default>) Failed to validate Schema: Schema-validation: wrong column type encountered in column [code] in table [`ComponentTypes`]; found [bpchar (Types#CHAR)], but expecting [varchar(255) (Types#VARCHAR)]
2026-03-11 13:54:36,358 ERROR [io.quarkus.hibernate.orm.runtime.schema.SchemaManagementIntegrator] (Hibernate post-boot validation thread for <default>) The following SQL may resolve the database issues, as generated by the Hibernate schema migration tool. WARNING: You must manually verify this SQL is correct, this is a best effort guess, do not copy/paste it without verifying that it does what you expect.

alter table if exists "ComponentTypes" alter column code set data type varchar(255);
alter table if exists "ComponentTypes" alter column description set data type varchar(255);
alter table if exists "ComponentTypes" alter column name set data type varchar(255);
.
.
.

Maybe you have a link/solution…?

Thanks

1. You should create a new schema and then add the tables to it. For example, add creating a new schema to db/create-tables-postgresql.sql and then create a new table in that schema:

CREATE SCHEMA IF NOT EXISTS retract;
CREATE TABLE IF NOT EXISTS retract."Countries" (
  ...
);

2. The error you see is coming from schema validation. By default, Quarkus/Hibernate ORM performs a post-startup validation of the database schema in dev mode and logs any mismatches. This behavior is controlled by: quarkus.hibernate-orm.validate-in-dev-mode (default: true). See this doc.

From your screenshot, it looks like the code column is defined as a fixed-length type in PostgreSQL (CHAR(2) / bpchar). In this case you have two options:

The first option (recommended):
Align the entity mapping with your DDL by explicitly specifying the column length, e.g.:

@Column(name = "code", length = 2)
private String code;

The second option:
Disable dev schema validation if you don’t want to keep entities strictly aligned with the DDL:
%dev.quarkus.hibernate-orm.validate-in-dev-mode=false

Hmm. Prefixing table names with their schema name is not good. One of my big tasks is to create DB scripts that can replicate the DB on any PG system. The remote development schema name is “retract.e”, so I will live with the default schema name “public” for dev setup.

As for the 2nd I set the property to false for now until I might add the meta info per column later.

Thanks so much for helping! Invaluable info.

Thanks for the update.

Prefixing table names with their schema name is not good. 

I understand that it may be not the best solution for all cases. In your case, the SET search_path command might be useful, for example:

CREATE SCHEMA IF NOT EXISTS retract;
SET search_path TO retract, public;
CREATE TABLE IF NOT EXISTS "Countries" (...);

In this case, table creation/search will be performed in the retract schema, and if it is not found, then in public.
This will allow you to avoid changing the script in multiple places when switching the prefix to another one, and instead define the schema only once in your initialization script.
Please feel free to correct me if I did not fully understand your scenario correctly.