IV62853: NCP_MODEL HANGS CAUSING OQL QUERIES TO TIMEOUT AND KEEPING FURTHER TOPOLOGY UPDATES FROM BEING PROCESSED

4.1.1-TIV-ITNMIP-Linux-FP0002
4.1.1_README_NLS_FP0002
4.1.1_README_FP0002
3.9.0-TIV-ITNMIP-Solaris-FP0005
3.9.0_INSTALL_NLS_FP0005
3.9.0_INSTALL_FP0005
3.9.0-TIV-ITNMIP-Windows-FP0005
3.9.0-TIV-ITNMIP-zLinux-FP0005
3.9.0-TIV-ITNMIP-Linux-FP0005
3.9.0-TIV-ITNMIP-AIX-FP0005
4.1.1_README_NLS_FP0001
4.1.1_README_FP0001
4.1.1-TIV-ITNMIP-Linux-FP0001
3.9.0.4-TIV-ITNMIP-CORE-OpenSSL-Windows-IF0001
3.9.0.4-TIV-ITNMIP-GUI-Windows-IF0001
3.9.0.4-TIV-ITNMIP-GUI-Window-IF0001
3.9.0.4-TIV-ITNMIP-GUI-Unix-IF0001
3.9.0.4-TIV-ITNMIP-CORE-OpenSSL-zLinux-IF0001
3.9.0.4-TIV-ITNMIP-CORE-OpenSSL-Solaris-IF0001
3.9.0.4-TIV-ITNMIP-CORE-OpenSSL-Linux-IF0001
3.9.0.4-TIV-ITNMIP-CORE-OpenSSL-AIX-IF0001
3.9.0.4-TIV-ITNMIP-zLinux-CORE-IF0001
3.9.0.4-TIV-ITNMIP-Windows-CORE-IF0001
3.9.0.4-TIV-ITNMIP-Solaris-CORE-IF0001
3.9.0.4-TIV-ITNMIP-Linux-CORE-IF0001
3.9.0.4-TIV-ITNMIP-AIX-CORE-IF0001
3.9.0.4_README_NLS_IF0001
3.9.0.4_README_GUI_IF0001
3.9.0.4_README_CORE_OpenSSL_IF0001
3.9.0.4_README_CORE_IF0001
IBM Tivoli Network Manager IP 3.9.0.4 Interim Fix 1, 3.9.0.4-TIV-ITNMIP-IF0001
IBM Tivoli Network Manager IP Edition 3.9.0 Fix Pack 5, 3.9.0-TIV-ITNMIP-FP0005
IBM Tivoli Network Manager IP Edition 4.1.1 Fix Pack 2, 4.1.1-TIV-ITNMIP-FP0002

APAR status

• Closed as program error.

Error description

• Multiple simulataneous updates to the topology db causes an
  internal ncp_model deadlock
  

Local fix

• restart ncp_model process
  

Problem summary

• ****************************************************************
  * USERS AFFECTED:                                              *
  * All users                                                    *
  ****************************************************************
  * PROBLEM DESCRIPTION:                                         *
  * ncp_model hangs causing OQL queries to timeout and keeping   *
  * further topology updates from being processed.               *
  ****************************************************************
  * RECOMMENDATION:                                              *
  * The following fixpacks will contain the fix:                 *
  * | fix pack | 3.9.0-ITNMIP-FP0005                             *
  * | fix pack | 4.1.1-ITNMIP-FP0001                             *
  * | fix pack | 3.9.4-ITNMIP-IF0001                             *
  ****************************************************************
  

Problem conclusion

• Multiple simultaneous updates to the topology db causes an
  internal ncp_model deadlock.
  The problem is that the m_SuspensionCount member variable in
  CMdlManagedStatus is not protected by any mutex. This means that
  multiple threads could all attempt to increment it at the same
  time and so it could therefore reach a value greater than 1
  before it had actually suspended the managed status polling
  thread (which it does only once if the count is at 1).
  
  Code changed to have the m_SuspensionCount protected when
  updated by any mutex.
  

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

  TIV NETWK MGR I

• Fixed component ID

  5724S4500

Applicable component levels

• R390 PSN

     UP

• R390 PSY

     UP

• R411 PSN

     UP

• R411 PSY

     UP