Yoann Rodière

Ok, then I can't reproduce this. So: > What logs do you see when you execute your reproducer? > > What does your .testcontainers.properties file look like? > > What's...

Is that the whole log? I don't see the [warning I mentioned earlier about reuse needing to be enabled explicitly](https://github.com/quarkusio/quarkus/issues/38687#issuecomment-1935759434), that's suspicious. Also: > `2024-02-21 10:42:33,634 INFO [org.tes.doc.DockerClientProviderStrategy] (build-58) Loaded...

Hello @m-stramel , kind reminder that we don't have enough information to investigate this problem (can't reproduce it, seemingly incomplete logs).

~~Upstream issue: https://hibernate.atlassian.net/browse/HHH-17559~~ Actually https://hibernate.atlassian.net/browse/HHH-17571

I didn't try this locally yet (have to download an Oracle container on a low-bandwidth connection...), but IIUC the main problem is that the interceptor doesn't trigger a rollback before...

> My guess is that Oracle is the only database provider that behaves like this. That's for sure, and from what I can see it's well-known to Oracle users; found...

> and threads are killed if they take too long to finish That's the thing: if threads are killed, the call to `Thread.sleep` should throw an `InterruptedException`, and if you...

Unfortunately I didn't have time to make progress on this. > And all that should happen before the DB connection gets closed, because why on earth would Quarkus stop DB...

> Maybe killing the applications threads don't cause an explicit rollback of the transaction? If you used `@Transactional(rollbackOn={Exception.class})` (and you said earlier you did try that), then `InterruptedException` definitely should...

Hey @segersb , while working on something else I stumbled upon this: https://quarkus.io/guides/lifecycle#graceful-shutdown Could you confirm that when you talk about graceful shutdowns, you mean it in this specific sense...