JR61640: UNATTENDED CLUSTER FAILURE MIGHT CAUSE SUBSEQUENT CONNECTION FAILURES FROM PX CONDUCTOR NODE TO PXYARN CLIENT.

IBM InfoSphere Information Server, Version 11.7.1.1 (Fix Pack 1)
IBM InfoSphere Information Server, Version 11.7.1.1 for AIX
IBM InfoSphere Information Server, Version 11.7.1.1 for Windows
IBM InfoSphere Information Server, Version 11.7.1.1 for Linux

APAR status

• Closed as fixed if next.

Error description

• A Hadoop  cluster failure can leave bad data in temporary files
  which will block YARN client connections even after subsequent
  reboots.
  Users will see connection failures from the PX conductor node to
   the PXYarn    client when the PXYarn client starts.
  
  
  There are two files which can cause this problem:
  
  
   /tmp/yarn_client_port.out or
  /tmp/yarn_client_port.$DSUSER.out
  
  Job logs show an error message like :
  ----
  Message Id: IIS-DSEE-TFPM-00464
  Message: main_program: Fatal Error: Could not get the address
  info to connect to the Application Master: Success.
  ----
  

Local fix

• Remove from Conductor node the offending temporary file :
  ----
  /tmp/yarn_client_port.out
  ----
   or :
  ----
  /tmp/yarn_client_port.$DSUSER.out
  ----
  and restart YARN Client
  

Problem summary

• A Hadoop  cluster failure can leave bad data in temporary files
  which will block YARN client connections even after subsequent
  reboots.
  Users will see connection failures from the PX conductor node to
  the PXYarn    client when the PXYarn client starts.
  There are two files which can cause this problem:
  /tmp/yarn_client_port.out
  or
  /tmp/yarn_client_port.$DSUSER.out
  

Problem conclusion

Temporary fix

• /tmp/yarn_client_port.out
  or :
  /tmp/yarn_client_port.$DSUSER.out
  
  and restart YARN Client
  

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

Applicable component levels

• RB71 PSY

     UP