IT27924: A race condition to close a TCP/IP socket exists which could cause TIME_WAIT status on the server

APAR status

• Closed as program error.

Error description

• A race condition exists between IBM MQ clients and queue manager
  when closing down a TCP/IP network connection.
  
  The network resource at the end which issues the close first
  enters a TIME_WAIT state for a period of time in which delayed
  TCP packets will be discarded. If the queue manager side is the
  first to close the network resource across many connections
  within a short period of time, this state could eventually lead
  to a shortage of available ephemeral ports to accept new
  connections.
  

Local fix

• Clients using SHARECNV(0) have a greater chance of the client
  closing the network connection first and so a TIME_WAIT on the
  server is usually avoided, but can't be guaranteed.
  

Problem summary

• ****************************************************************
  USERS AFFECTED:
  Users of MQ TCP/IP client connections are affected.
  
  
  Platforms affected:
  MultiPlatform
  
  ****************************************************************
  PROBLEM DESCRIPTION:
  A race condition exists when sending and receiving the final MQ
  protocol flows to indicate the end of a TCP/IP connection. The
  first side of a channel that closes the network connection will
  cause the ephemeral port that was used for the connection to
  enter a TIME_WAIT state for a short period of time to prevent
  its reuse. The duration of the TIME_WAIT state is relative to
  the maximum segment lifetime, this is to prevent delayed TCP
  packets from interfering with a new conversation reusing the
  same destination port.
  
  As a port in the TIME_WAIT state may be unavailable for reuse
  for a few minutes, it is desirable for MQ clients to be the side
  of the connection to close the network resources, this is so
  that the the range of ephemeral ports available for new
  connections is not diminished on the network interface being
  used by the queue manager.
  

Problem conclusion

• Multiplexed Java/JMS, .NET and native clients have been altered
  to close network resources immediately after sending the
  protocol flow to the queue manager to close the last connection.
  Responding message channel agents (MCAs) have been modified to
  wait briefly for the connection to be closed by the other end of
  the channel when it is deemed appropriate to do so. Under normal
  circumstances these changes cause TCP/IP TIME_WAIT state to
  occur on the initiating end of the channel.
  
  ---------------------------------------------------------------
  The fix is targeted for delivery in the following PTFs:
  
  Version    Maintenance Level
  v8.0       8.0.0.13
  v9.0 LTS   9.0.0.8
  v9.1 CD    9.1.3
  v9.1 LTS   9.1.0.3
  
  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

  IBM MQ BASE M/P

• Fixed component ID

  5724H7261

Applicable component levels

• R904 PSY

     UP