JIRA link: https://issues.redhat.com/browse/QUARKUS-2492
Programmatic distributed JDBC transaction API based on Narayana library
Documentation: Narayana Documentation Quarkus JTA
Test development will focus on
- Quarkus JDBC transactions based on programmatic approach
- Quarkus JDBC transactions based on lambda approach
- Verify commons transactions states as "begin", "commit" and "rollback"
- Reactive flavor is out of scope, currently only classic approach is supported
- The following databased will be covered: Oracle, MySQL, MariaDB, Mssql, PostgreSQL
- Areas like OpenAPI, Tracing and Metrics were considered when preparing the test plan
- Closer integration for Tracing and Metrics topic will be evaluated and brought to the development team
- TD will cover custom metric handled by the application code and ensure true transactional behaviour - e.g. rollback
We are going to create a new module under SQL/DB folder, in order to develop new JDBC transaction scenarios. The main idea is to develop a "real scenario" where the transactions are handler in a service layer, rather than in entity level. This new module is going to increase the compile time, especially in native mode, and is important to run the scenario in all the platforms, OCP, Baremetal (JVM and native mode). I would say that each native compilation could takes about 3 min and then 1 extra minute in order to run the scenario itself. Overall my prediction is that the total amount of time required to run transaction API scenarios is about 30 min.
Following actions were taken to ensure familiarity:
- Ensure documentation is clear and easy to understand
- Ensure good user experience and simplicity of use
- Review upstream coverage
Narayana JTA transaction support several thread semactics such as "DISALLOW_EXISTING", "JOIN_EXISTING", "REQUIRE_NEW", "SUSPEND_EXISTING". Currently we are just covering the default one (requireNew), but would be great in the future to cover the others.
- Tester: Pablo Gonzalez Granados [email protected]