Saravanan Subiramaniam created GROOVY-11729:
-----------------------------------------------

             Summary: Increase in heap memory caused by 
org.codehaus.groovy.vmplugin.v8.CacheableCallSite.latestHitMethodHandleWrapperSoftReference
                 Key: GROOVY-11729
                 URL: https://issues.apache.org/jira/browse/GROOVY-11729
             Project: Groovy
          Issue Type: Question
    Affects Versions: 4.0.24
            Reporter: Saravanan Subiramaniam


We recently upgraded Groovy from 3.x to 4.0.24 in our application and noticed 
increase in heap memory in customer environments. For example, out of 2.3 GB 
heap space, 1.2 GB is occupied by RepeatableWriteUnitOfWork - which is a 
transactional stuff from EclipseLink. This issue is similar to GROOVY-10773. 
We've 66K org.codehaus.groovy.vmplugin.v8.CacheableCallSite objects and the 
total retained set of org.codehaus.groovy.vmplugin.v8.CacheableCallSite is 1.2 
GB. Please see the paths to GC root (with all references) below:
{code:java}
class java.lang.invoke.LambdaForm$MH @ 0x74b360088 (System Class)
 └── <resolved_references> java.lang.Object[6] @ 0x74b3600f8
     └── [1] java.lang.invoke.BoundMethodHandle$Species_LLLLL @ 0x74b363940
         └── argL4 java.lang.invoke.BoundMethodHandle$Species_LL @ 0x74b363c58
             └── argL1 class 
mycompany.dynamiclogic.MY_DYLO$_getDuplicateDetailsForPerson_closure2 @ 
0x77226a958
                 └── <resolved_references> java.lang.Object[172] @ 0x77591fd68
                     └── [66] org.codehaus.groovy.vmplugin.v8.CacheableCallSite 
@ 0x775930328
                         └── latestHitMethodHandleWrapperSoftReference 
java.lang.ref.SoftReference @ 0x789b7bd68
                             └── referent 
org.codehaus.groovy.vmplugin.v8.MethodHandleWrapper @ 0x789b7bd90
                                 └── cachedMethodHandle 
java.lang.invoke.BoundMethodHandle$Species_LLLL @ 0x789b7bdb0
                                     └── argL1 
java.lang.invoke.MethodHandleImpl$CountingWrapper @ 0x789b7be20
                                         └── target 
java.lang.invoke.BoundMethodHandle$Species_LLLL @ 0x789b7be58
                                             └── argL1 
java.lang.invoke.MethodHandleImpl$CountingWrapper @ 0x789b7beb0
                                                 └── target 
java.lang.invoke.BoundMethodHandle$Species_LLLL @ 0x789b7bee8
                                                     └── argL0 
java.lang.invoke.BoundMethodHandle$Species_LL @ 0x789b7bf18
                                                         └── argL1 
com.mycompancy.dynlogic.components.GroovyAwareList$GroovyAwareListMetaClass @ 
0x789b7c0c0
                                                             └── this$0 
com.mycompancy.common.dynlogic.components.GroovyAwareList @ 0x789b7bf58
                                                                 └── 
BasePersistable com.mycompancy.entities.MyEntityDomain @ 0x789b7c70c
                                                                     └── 
_persistence_listener 
com.mycompancy.support.eclipselink.MyAttributeChangeListener @ 0x789b7f958
                                                                         └── 
uow org.eclipse.persistence.internal.sessions.RepeatableWriteUnitOfWork @ 
0x781c25650 {code}
Why is the returned value from method is cached in 
{*}CacheableCallSite.latestHitMethodHandleWrapperSoftReference{*}?

Although it is SoftReference, these objects survive Full GC and the heap memory 
is heavily occupied by these objects.  We've experienced that K8s kills the POD 
as it reaches the memory limit and my guess is that the increase in heap space 
contributes to the issue. Is there a way to skip caching the returned value 
from the methods (or change it to WeakReference instead)?

We did not run into this issue with Groovy 3. So, looking forward to 
suggestions on how to address this/keep the memory consumption low.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to