[Bug]: Race condition between DB init and test executions from JdbcDatabaseContainer based containers about testcontainers-java

[kiview]

Thanks for reporting and the detailed analysis @luanhui0420.

It was found out that the JdbcDatabaseContainer overrides the waitUntilContainerStarted and writes its own DB connection based logic and ignore the waitingStrategy passed in either via constructor or by .waitingFor.

This is indeed the case for most subclasses of JdbcDatabaseContainer. And I imagine, depending on the internal image lifecycle, the same issue of wait behavior being independent of init script execution (hence introducing a race condition), could be present for other database container implementations as well.

[luanhui0420]

Thanks for reporting and the detailed analysis @luanhui0420.

It was found out that the JdbcDatabaseContainer overrides the waitUntilContainerStarted and writes its own DB connection based logic and ignore the waitingStrategy passed in either via constructor or by .waitingFor.

This is indeed the case for most subclasses of JdbcDatabaseContainer. And I imagine, depending on the internal image lifecycle, the same issue of wait behavior being independent of init script execution (hence introducing a race condition), could be present for other database container implementations as well.

I agree it impacts all the subclasses. It is usually not noticeable if the init script is short as well.

[tejaslodayadd]

@kiview do you suggest any workaround for this? Thanks