IV26784: UA:LINUX ONLY, UA STOPS MONITORING SNMP TRAPS WHEN CHECKSUM ERROR RECEIVED

IBM Tivoli Monitoring 6.2.3 Fix Pack 5 (6.2.3-TIV-ITM-FP0005)
IBM Tivoli Monitoring 6.2.3 Fix Pack 4 (6.2.3-TIV-ITM-FP0004)

APAR status

• Closed as program error.

Error description

• When UA received a checksum error in the packet received from
  snmp trap, kumstrap exits automatically.
  
  kumstrap.log entries:
  ----------------------
  50305C51 ----- Trap socket select completed. nfds=1
  50305C51 ----- Checking trapSSE ReadMasks
  50305C51 ----- Clearing ReadMask for IPv4 socket 4
  50305C51 ----- Socket 4 selected for UDP data, addrlen 16
  50305C51 ***** trap receive error on socket 4, errno 11 addrlen
  16
  
  
  kumstrap exit:
  ----------------------
  13:24:01.516639 recvfrom(4, 0x7fff7d868600, 4000, 0,
  0x7fff7d868180, 0x7fff7d86817c) = -1 EAGAIN (Resource
  temporarily unavailable)
  13:24:01.536794 write(3, "50305C15 ----- Socket 4 added to"...,
  324) = 324
  13:24:01.536981 close(3)                = 0
  13:24:01.537078 munmap(0x7f5283a7d000, 4096) = 0
  13:24:01.548795 exit_group(0)           = ?
  
  
  Modify the UA code for Linux only to suppress the 'exit
  logic' and instead to reissue the socket select() call.
  

Local fix

Problem summary

• Universal Agent on LINUX only, stops monitoring SNMP traps when
  Checksum error received
  
  
  When software that emits SNMP trap injects CheckSum error in the
  trap data, Linux O/S socket select() API does not cope with this
  circumstance, falsely returning success status on 'select'.
  Subsequent socket recvfrm() API fails to receive data since
  there is no valid data.
  
  Universal Agent SNMP trap 'listening' code is a continuous loop,
  thus implemented design was to exit from the loop on recvfrm()
  failure and terminate the SNMP trap 'listening' thread;
  otherwise Universal Agent could potentially fall into a tight,
  infinite loop of repeatedly issuing recvrm() calls that return
  failed status.
  

Problem conclusion

• Customer resolved problem by correcting CheckSum error in their
  Perl script.
  
  APAR fix only delivers nominal enhancement to Universal Agents's
  reaction to bad data arriving on SNMP trap listening socket.
  This APAR "fix" added a counter that checks for 5 consecutive
  recvfrm() failures and if so then original Universal Agent
  behavior of exiting from loop on recvfrm() failure and
  terminating SNMP trap 'listening' thread is carried out -  this
  APAR can not fix a customer's code.
  
  The fix for this APAR is included in the following maintenance
  vehicle:
  
     | fix pack | 6.2.3-TIV-ITM-FP0003
  
  Note: Search the IBM technical support web site for maintenance
  package availability.
  

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

  UNIVERSAL AGENT

• Fixed component ID

  5724K1000

Applicable component levels

• R623 PSY

     UP