For OOM please look carefully at the OOM message from DR as it contains codes which describe what kind of OOM it is. Without knowing the kind, blindly increasing limits may not do anything. Even once known, there are multiple types of limits that can lead to that failure (as Kirill pointed out).
For the heap, the initial reservation (8G by default for 64-bit) is *not* a hard limit and allocation will continue past that point (though there are possible performance issues: issue #2115), so unless the application is taking up all of the address space (it's possible for 32-bit) you are not going to see OOM involving the heap address space. Thus, increasing -vmheap_size is probably not going to accomplish anything toward eliminating OOM. You might see a commit limit hit on a heap allocation coming from system overcommit settings or mmap limits (again see Kirill's thread).