In my production environment, the ES cluster frequently encounters data too large exception.
When I dumped the heap memory and analyzed it through MAT, I found the following confusing points:
- There are a large number of unreachable objects
May I ask when this part of memory can be recycled and what is stored there?