Why does quarkus use hibernate reactive to operate the database and the query method does not use database connection pool

Question

I use the quarkus framework, integrate hibernate reactive, and use the panache list method to query. I find that it do not use the connection pool, but use the connection directly. As a result, the query cannot be used after the connection is closed.

SqlClientConnection

public CompletionStage<RowSet<Row>> preparedQuery(String sql, Tuple parameters) {
        feedback(sql);
        String processedSql = usePostgresStyleParameters ? Parameters.process( sql, parameters.size() ) : sql;
        return Handlers.toCompletionStage(
                        handler -> client().preparedQuery( processedSql ).execute( parameters, handler )
        );
}

PanacheQueryImpl

  @Override
    public <T extends Entity> Uni<List<T>> list() {
        return delegate.list();
    }

I'd like to know how to use pool when querying?

Answer 1

Technically it doesn't need a traditional connection pool, because all operations are non-blocking, so you can easily saturate your CPU completely with a single connection to the database.

In practice the default is to use a single connection for each Vert.x context, and by default you'll have two such contexts per CPU.

When you say "query cannot be used after the connection is closed" I suspect you're actually hitting a bug we fixed recently; please add more details to a new question; I suspect the solution for your issue (in that case) is that you need to use Panache Reactive from a non-blocking context, such as using RESTEasy Reactive for your endpoint.