IT18394: MQ channel process amqmrppa or amqcrsta consumes high cpu on Windows systems

APAR status

• Closed as program error.

Error description

• The process status output for the MQ channel process(amqrmppa or
  runmqchl or amqcrsta)  shows
  high CPU or 100% CPU usage for the channel process. MQ trace of
  channel process(amqrmppa/runmqchl/amqcrsta)
  shows a tight loop with repeated send calls failing with the
  error WSAECONNRESET(10054 (0x2746)).
  
  006EF815 14:21:53.457939   4964.9       CONN:971534 -------{
  ccigsk_secure_soc_write
  006EF816 14:21:53.457945   4964.9       CONN:971534 Channel
  Name:TESTCHANNEL       Local:127.0.0.1(1414)
  PeerName:127.0.0.1(55411)
  006EF817 14:21:53.457953   4964.9       CONN:971534
  gsk_secure_soc_write: input: gsk_soc_handle=0x0000000006BB27D0,
  
  pBytesToWrite=0x000000000784E730,nBytesToWrite=0,pnBytesWritten=
  0x000000000784E688
  006EF818 14:21:53.457988   4964.9       CONN:971534 --------{
  cciTcpSslWriteCallback
  006EF819 14:21:53.457996   4964.9       CONN:971534
  cciTcpSslWriteCallback: input: nBytesToWrite=15397
  006EF81A 14:21:53.458002   4964.9       CONN:971534 ---------{
   cciTcpSend
  006EF81B 14:21:53.458007   4964.9       CONN:971534 Sending
  15397 bytes from 0000000007623E30
  006EF81C 14:21:53.458013   4964.9       CONN:971534 ----------{
   send
  006EF81D 14:21:53.458025   4964.9       CONN:971534 ----------}!
  send (rc=Unknown(2746))
  006EF81E 14:21:53.458041   4964.9       CONN:971534 Data:
  0xffffffff
  006EF81F 14:21:53.458053   4964.9       CONN:971534 Sent -1
  bytes
  006EF820 14:21:53.458065   4964.9       CONN:971534 ----------{
   cciTcpGetNameandAddress
  006EF821 14:21:53.458071   4964.9       CONN:971534 Reverse DNS
  lookups: FALSE
  006EF822 14:21:53.458078   4964.9       CONN:971534 Resolved
  name: 127.0.0.1
  006EF823 14:21:53.458085   4964.9       CONN:971534 ----------}
   cciTcpGetNameandAddress (rc=OK)
  006EF824 14:21:53.458089   4964.9       CONN:971534 RetCode
  (rrcE_SEND_FAILED)
  006EF825 14:21:53.458095   4964.9       CONN:971534 ---------}!
  cciTcpSend (rc=rrcE_SEND_FAILED)
  006EF826 14:21:53.458102   4964.9       CONN:971534 --------}!
  cciTcpSslWriteCallback (rc=Unknown(FFFFFFFF))
  006EF827 14:21:53.458202   4964.9       CONN:971534
  gsk_secure_soc_write: output: gsk_soc_handle=0x0000000006BB27D0,
  
  pBytesToWrite=0x000000000784E730,nBytesToWrite=0,pnBytesWritten=
  0x000000000784E688, nBytesWritten=0
  006EF828 14:21:53.458212   4964.9       CONN:971534 -------}!
  ccigsk_secure_soc_write (rc=Unknown(227))
  006EF829 14:21:53.458225   4964.9       CONN:971534 -------{
  xcsWaitFd
  006EF82A 14:21:53.458231   4964.9       CONN:971534 fdcount=1,
  fd[0]=1048 option=2 timeout=(0sec, 0usec)
  006EF82B 14:21:53.458238   4964.9       CONN:971534 timeout=20s
  (+0us)
  006EF82C 14:21:53.458245   4964.9       CONN:971534 --------{
  select
  006EF82D 14:21:53.458258   4964.9       CONN:971534 --------}!
  select (rc=Unknown(1))
  

Local fix

• Restart the channel(s) being serviced by the affected process.
  

Problem summary

• ****************************************************************
  USERS AFFECTED:
  Users using MQ SSL channels on Windows systems.
  
  
  Platforms affected:
  Windows
  
  ****************************************************************
  PROBLEM DESCRIPTION:
  The MQ channel agent sending code did not correctly handle the
  case where the Windows socket reported the error code EWOULDBLK
  to report a saturated send buffer followed by the error code
  ECONNRESET to report a broken connection. This caused the
  channel process to enter a tight loop, instead of reporting the
  broken connection as expected.
  

Problem conclusion

• The MQ channel agent logic has been modified to correctly handle
  this connection failure.
  
  ---------------------------------------------------------------
  The fix is targeted for delivery in the following PTFs:
  
  Version    Maintenance Level
  v8.0       8.0.0.7
  v9.0 CD    9.0.2
  v9.0 LTS   9.0.0.1
  
  The latest available maintenance can be obtained from
  'WebSphere MQ Recommended Fixes'
  http://www-1.ibm.com/support/docview.wss?rs=171&uid=swg27006037
  
  If the maintenance level is not yet available information on
  its planned availability can be found in 'WebSphere MQ
  Planned Maintenance Release Dates'
  http://www-1.ibm.com/support/docview.wss?rs=171&uid=swg27006309
  ---------------------------------------------------------------
  

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

  WMQ BASE MULTIP

• Fixed component ID

  5724H7251

Applicable component levels

• R800 PSY

     UP