JR43259: Normal TCP errors CAUSE INFOMATION SERVER 8.5 TO CREATE MULTIPLE CORE FILES ON LINUX

APAR status

• Closed as program error.

Error description

• Problem Description:
  If an I/O error occurs the PX section leader process immediately
  closed its side of the communication socket with the conductor
  process. This caused the conductor process to to fail and dump
  core. Over thousands of job runs with MQ or z/OS stages,
  occasional communication errors would lead to hundreds of core
  file on the PX server machine, which filled up the file system.
  The fix checks for several common TCP errors so that the PX
  shuts down normally without creating a core.
  

Local fix

Problem summary

• Information Server is creating core files regularly and filling
  up disk space.
  
  Customer believes they have identified they are related to lost
  connection, either MQ Server channel (as in these files) or a
  z/OS DB2 connection.
  
  Using gdb on core file shows:
  
  Program terminated with signal 11, Segmentation fault.
  #0  0x00002b9d8f8646a9 in APT_PMMessagePort::makeMessage(char*,
  int) ()
     from
  /opt/IBM/InformationServer/Server/PXEngine/lib/liborchx86_64.so
  

Problem conclusion

• This is a very low incidence problem.
  
  
  
                      Solution:
  In case of I/O error, we are deactivating Section leader,
  causing section leader marked as not
  usable in this step. During this process, the control port
  (conductor side of the section leader)
  removed from polling group and further dumping core.
  Socket errors causing the msgport to set to state bad and
  causing the core dump.
  Check was added to avoid the above cases.
  

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

  WIS DATASTAGE

• Fixed component ID

  5724Q36DS

Applicable component levels

• R850 PSY

     UP

• R870 PSY

     UP