JR34050: OUTOFMEMORY ERRORS WHEN RESUBMITTING LARGE NUMBER OF FAILED EVENTS IN WPS 6.0.2.5

WebSphere Enterprise Service Bus V6.2.0 Fix Pack 2 (6.2.0.2)
WebSphere Process Server V6.2.0 Fix Pack 2 (V6.2.0.2)
WebSphere Enterprise Service Bus V6.1.0 Fix Pack 4 (6.1.0.4)
WebSphere Process Server V6.1.0 Fix Pack 4 (6.1.0.4)

APAR status

• Closed as program error.

Error description

• OutOfMemory error occurs in WebSphere Process Server when large
  number of failed events are resubmitted. The heapdumps indicate
  a leak in Workspace component as below:
  
  1,788,987,048 (83%) [312] 1 class
  com/ibm/ws/sm/workspace/WorkSpaceManagerFactory 0x507c6960
   1,788,986,736 (83%) [40] 4
  com/ibm/ws/sm/workspace/impl/WorkSpaceManagerImpl 0x581a9710
    1,788,986,680 (83%) [32] 2
  java/util/Collections$SynchronizedMap
  0x581a5638
     1,788,986,648 (83%) [48] 1 java/util/HashMap 0x581a5658
      1,788,986,600 (83%) [4,112] 356 array of
  java/util/HashMap$Entry
  0x92c174b0
       16,293,408 (0%) [32] 3 java/util/HashMap$Entry 0xb8129970
       16,293,408 (0%) [32] 3 java/util/HashMap$Entry 0x7d7807a0
       12,220,344 (0%) [32] 3 java/util/HashMap$Entry 0x85ee4b08
       12,220,344 (0%) [32] 3 java/util/HashMap$Entry 0x8ca39738
  
  The memory leak is induced when Failed Event Manager code calls
  the below methods:
  
  at java.security.AccessController.doPrivileged1(Native Method)
  at
  java.security.AccessController.doPrivileged(AccessController.jav
  a(Compiled Code))
  at
  com.ibm.wbiserver.manualrecovery.util.FailedEventMessageUtil.res
  olveWASVariable(FailedEventMessageUtil.java:1211)
  at
  com.ibm.wbiserver.manualrecovery.util.FailedEventMessageUtil.get
  TargetSignificance(FailedEventMessageUtil.java:1166)
  
  There is a known memory leak issue aroung getTargetSignificance
  which is resolved for SCA code, but seems Failed Event Manager
  code is not modified to include this fix
  

Local fix

• Clearing the wstemp directories will provide temporary relief
  

Problem summary

• ****************************************************************
  * USERS AFFECTED:  Websphere Process Server 6.0.2 and 6.2.0.1  *
  *                  users who are resubmitting a large number   *
  *                  of failed events.                           *
  ****************************************************************
  * PROBLEM DESCRIPTION: When a large number of failed events    *
  *                      are resubmitted, WebSphere              *
  *                      Process Server runs out of memory.      *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  When a large number of failed events
  are resubmitted, the recovery component has a memory leak
  which causes the server to run out of memory
  

Problem conclusion

• The memory leak is fixed by destroying the session once the
  call is completed. This has been fixed in WebSphere Process
  Server v6.0.2.5. and v6.2.0.1.
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

• APAR is sysrouted TO one or more of the following:

Fix information

• Fixed component name

  WEB PROC SERVER

• Fixed component ID

  5724L0100

Applicable component levels

• R602 PSY

     UP

• R620 PSY

     UP