JProfiler: trying to find memory leak

My application needs about 10GB of RAM for a specific input, where for normal inputs, about 1GB is sufficient. A closer analysis with JProfiler shows that (after GC) quite a lot of memory is used by the standard classes from java.util.*

:

LinkedHashMap$Entry

, HashMap$Entry[]

, LinkedHashMap

, HashMap$KeySet

, HashMap$EntrySet

, LinkedHashSet

, TreeMap$Entry

And TreeMap

(in this order) and related classes. The following entry refers to a class in my own code where the number of instances and the amount of memory used seems to be very reasonable.

In detail, with a total heap of about 900 MB, I see the following entries Size

in the view All Objects

:

• LinkedHashMap$Entry
  
  : 418 MB
• HashMap$Entry[]
  
  : 178 MB
• LinkedHashMap
  
  : 124 MB
• HashMap$KeySet
  
  : 15 MB

The memory in use LinkedHashMap

seems to be too high, even considering that each is LinkedHashSet

supported LinkedHashMap

.

I recorded the selection of objects in JProfiler and watched Allocation Hot Spots

for LinkedHashMap

. There I see entries that I don't understand:

• The third entry shows a hotspot (with 6.5% memory allocated) named X.<init>
  
  where X
  
  is a class in my own code. The constructors for this method have nothing to do with LinkedHashMap
  
  . After this entry, Thread.run
  
  at the end, a slow decline is shown from 6.5% to 5.8% at Thread.run
  
  . What is the problem with my code in X
  
  ? Why is this shown here?
• About 8% of the allocated memory is mapped to a named hotspot java.util.HashSet.iterator
  
  . After this entry along the path with the highest percentage (first entry: 2.8%), I get several methods inside my code until finally it is shown java.lang.Thread.run
  
  (with 2.8%). What does it mean? As far as I know, the method Thread.run
  
  does not create instances LinkedHashMap
  
  . What is the relationship with the method iterator
  
  ?

In general, how do I find code that contains references to (many) LinkedHashMap

objects? Using Heap Walker, I can only see a lot of instances, but I can't see any pattern (even when watching the paths to the GC roots). In my experiments, all instances appear to be fine.

Possibly important things:

• My application creates a result (for further processing) and there is high memory for this construct. The design is constantly creating objects, so waiting for a stable point and then observing each created object is LinkedHashMap
  
  impossible.
• I have good computers for debugging (up to 48 cores and 192GB of RAM, maybe even more).
• java version "1.7.0_13" (Java (TM) SE Runtime (build 1.7.0_13-b20), Java HotSpot (TM) 64-bit Server VM (build 23.7-b01, mixed mode))
• JProfiler 7.2.2 (Build 7157) Licensed