IBM Support

PI72065: LOOP IN LIBERTY Z/OS SERVER WHEN ASYNCIO IS ENABLED.

Subscribe

You can track all active APARs for this component.

 

APAR status

  • Closed as program error.

Error description

  • After FixPack 16.0.0.3 which added Asynchronous TCP/IP
    Sockets
    I/O for Liberty on z/OS, a loop in native routine
    aio_exit_routine in mvs_aio_services.mc in load module
    bbgzsafm running under an SRB.
    This causes the system to be unresponsive by monopolizing
    each CPU.  Once all Processors are looping you cannot
    stop or cancel the Liberty Server.
    This new function must be enabled with a RACF profile
    specific
    to the new function and the Angel Process must also be
    running.
    Reference link:   https://ibm.biz/BdssKy
    You may see recursieve entries in the SYSTRACE output:
    0001-00BC 00000000 SSRV 120 81309210 00000000 0000261A
    ......
    .....D18C67AA TimerDIE
    There is a circumvention.
    The code that causes this problem is in a timeout routine.
    If you disable the timeout routine this problem won't occur.
    To disable related timeout routine set the following
    property
    in the bootstrap.properties:
    com.ibm.io.async.AIOHandlerCompletionTimeout=0
    

Local fix

  • The code that causes this problem is in a timeout routine.
    If you disable the timeout routine this problem won't occur.
    To disable related timeout routine set the following property
    in the bootstrap.properties:
    com.ibm.io.async.AIOHandlerCompletionTimeout=0
    

Problem summary

  • ****************************************************************
    * USERS AFFECTED:  All users of IBM WebSphere Application      *
    *                  Server Liberty for z/OS                     *
    ****************************************************************
    * PROBLEM DESCRIPTION: Loop in routine aio_exit_routine in     *
    *                      mvs_aio_services.mc in bbgzsafm running *
    *                      under an SRB.                           *
    ****************************************************************
    * RECOMMENDATION:                                              *
    ****************************************************************
    A timeout routine that processes stack elements had incorrect
    logic which lead to a circular stack.  The circular stack
    caused
    the loop while running under an SRB.  This lead to the server
    and the system becoming unresponsive.
    

Problem conclusion

  • Code has been modified in the SRB related routines to clear
    pointer values after a successful removal of a stack element.
    This will help prevent a circular stack.  Also, the logic
    within the timeout routine has been corrected to prevent
    chaining errors.
    
    The fix for this APAR is currently targeted for inclusion in fix
    pack 17.0.0.1.  Please refer to the Recommended Updates page for
    delivery information:
    http://www.ibm.com/support/docview.wss?rs=180&uid=swg27004980
    

Temporary fix

  • The Asynchronous I/O Feature (AsyncIO)  is disabled by default.
    It is enabled by granting read access for the server's user ID
    to the specific SAF profile named BBG.AUTHMOD.BBGZSAFM.ZOSAIO
    and having an Angel Process active for the Liberty Server.
    There are 2 ways to disable AsyncIO: 1) remove the read access
    from the SAF profile named BBG.AUTHMOD.BBGZSAFM.ZOSAIO; OR 2)
    add the following property to the bootstrap.properties file:
    com.ibm.ws.tcpchannel.useZosAio=false
    
    It is possible to avoid this problem completely while still
    continuing to use the Asynchronous I/O feature by setting the
    following property in bootstrap.properties:
    com.ibm.io.async.AIOHandlerCompletionTimeout=0
    The above property allows the Asynchronous I/O feature to be
    used but bypasses the specific code path that leads to the
    problem.
    

Comments

APAR Information

  • APAR number

    PI72065

  • Reported component name

    LIBERTY PROF -

  • Reported component ID

    5655W6514

  • Reported release

    CD0

  • Status

    CLOSED PER

  • PE

    NoPE

  • HIPER

    NoHIPER

  • Special Attention

    NoSpecatt / Xsystem

  • Submitted date

    2016-11-09

  • Closed date

    2017-01-03

  • Last modified date

    2017-01-03

  • 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

    LIBERTY PROF -

  • Fixed component ID

    5655W6514

Applicable component levels

  • RCD0 PSY

       UP

[{"Business Unit":{"code":"BU054","label":"Systems w\/TPS"},"Product":{"code":"SG19M","label":"APARs - z\/OS environment"},"Platform":[{"code":"PF054","label":"z Systems"}],"Version":"CD0","Line of Business":{"code":"","label":""}}]

Document Information

Modified date:
14 December 2020