PM22746: Link indexer of XHTML pages causes 100% CPU usage

APAR status

• Closed as program error.

Error description

• Environment:
  - Microsoft Windows XP SP3 or Windows 7
   - IBM Rational Software Architect for Websphere version
  8.0
     or
     IBM Rational Application Developer version 8.0
  
  Reproducible Y/N: Y
  
  Description of problem:
  After importing a project interchange file which contains 50 or
  more projects
  (which includes web projects with .xhtml files) into IBM
  Rational Application Developer
  verion8.0, the CPU usage may go very high (around 100% for
  single CPU and 50% for
  dual cores).
  Even after the full workspace build and validation have finished
  the CPU usage may stay high.
  
  Generating Javacore
  (following technote
  http://www.ibm.com/support/docview.wss?uid=swg21261089 )
  during the high CPU usage shows that the common stack in all the
  core files is the following:
  
  at java/util/zip/ZipFile.open(Native Method)
  at java/util/zip/ZipFile.<init>(Bytecode PC:98(Compiled Code))
  at java/util/zip/ZipFile.<init>(Bytecode PC:10(Compiled Code))
  at
  org/eclipse/wst/sse/core/internal/util/JarUtilities.getCachedInp
  utStream(Bytecode PC:10(Compiled Code))
  at
  org/eclipse/wst/sse/core/internal/util/JarUtilities.getInputStre
  am(Bytecode PC:51(Compiled Code))
  at
  org/eclipse/wst/sse/core/internal/util/JarUtilities.getInputStre
  am(Bytecode PC:72(Compiled Code))
  at
  org/eclipse/jst/jsp/core/taglib/ProjectDescription.restoreRefere
  nces(Bytecode PC:805(Compiled Code))
  at
  org/eclipse/jst/jsp/core/taglib/ProjectDescription.<init>(Byteco
  de PC:195)
  at
  org/eclipse/jst/jsp/core/taglib/TaglibIndex.createDescription(By
  tecode PC:70)
  
  A cache with a limited capacity causes a lot of meta data to be
  reloaded from disk repeatedly, which tends to be an expensive
  operation.
  
  Workarounds:
  
  * Exclude the XHTML files from link indexer by going to:
  Window > Preferences > Web > Links > add *.xhtml to ignore list
  > confirm and apply changes).
  OR
  * Start with a clean workspace and add '*' to the link indexing
  ignore rules.
  (This essentially turns off link indexing).
  Import all projects, wait for build and validation to complete
  Restart the product
  Remove the '*' indexing ignore rule and OK the Links preference
  page.
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED:                                              *
  ****************************************************************
  * PROBLEM DESCRIPTION:                                         *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  In Rational Application Developer 8, the index used to
  maintain tag library information is not retaining enough
  data in its cache. Reparsing the JAR files looking for the
  correct tag library definitions is causing high CPU usage.
  

Problem conclusion

• The tag library index was reworked so that its information
  isn't purged from the cache at such a high frequency.
  

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

  RATL APP DEV WI

• Fixed component ID

  5724J1901

Applicable component levels

• R800 PSN

     UP