I am having this issue as well.

It's really frustrating that we have to choose between @OneToOne lazy loading or persist cascade.

Our team spent several hours on this issue, and were so glad to finally see the issue here.

But's it's really a bummer that the issue has lied here for more than two years...