Odtnhj

The following is seen in a hs_err_pid####.log file.
The eden space is full as well as the old space, but the from/to space is at 0%

Heap:
 PSYoungGen total 197,652K, used 190,955K [0x00000000eab00000, 0x00000000f8000000, 0x0000000100000000)
 eden space 190,956K, 99% used [0x00000000eab00000,0x00000000f657fff8,0x00000000f6580000)
 from space 6,646K, 0% used [0x00000000f7080000,0x00000000f7080000,0x00000000f7700000)
 to space 11,364K, 0% used [0x00000000f6580000,0x00000000f6580000,0x00000000f7080000)
 ParOldGen total 699,397K, used 699,341K [0x00000000c0000000, 0x00000000eab00000, 0x00000000eab00000)
 object space 699,332K, 99% used [0x00000000c0000000,0x00000000eaaffff8,0x00000000eab00000)
 Metaspace used 71522K, capacity 76253K, committed 129276K, reserved 1126500K
 class space used 7619K, capacity 11188K, committed 52466K, reserved 1048576K

Could this occur when dealing with LARGE Arrays, HashTable or whatever?
Is it possible that when a minor GC occurs, the from/to space is too small to handle the object size and as a result it goes straight to the old space?

Has there been a common cause that people have encountered when seeing this type of behavior?

I would suggest to run a Eclipse MAT or similar tool and check which object is taking up the memory.

I had faced a similar issue in Kafka and it was due to a memory leak related to managed bean. In my case the Beans were very small in size, but a lot in number.

So your Old gen can get filled up regardless of the size of the object, as long as JVM thinks it needs to be kept alive.