PK65508: HANDLING OF UN-RESPONSIVE MASES

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• Message EYUCL0122E is issued when a dump is taken.  The dump
  shows COIR has an XLST task with a stack which indicates that
  there are no free SMTBs on the CLSD_FRE_SMTB chain.  The CMAS
  trace showed an EXCEPTION for COIR with trace point id 8 out of
  CTBT.  The pointID 8 is CMPI_NO_ELE.  Out of 2048 SMTBs
  available to the CMAS, we have a around 1900 so we are not
  running out of SMTBs.
  .
  The MASes on the LPAR were holding back the CMAS from completing
  work in a timely manner, which either can cause the CMAS to have
  problems to have problems with other CMASes or just with the
  local MASes.
  This apar will put out warning messages about any possible
  un-responsive MASes so preventive measures can be taken to avoid
  further problems.
  .
  Additional Keywords:
  unresponsive nonresponsive un-responsive non-responsive
  mas
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED: All CICSPlex/SM V3R1M0 Users                 *
  ****************************************************************
  * PROBLEM DESCRIPTION: Since a single non-responsive MAS can   *
  *                      result in a CMAS not being able to      *
  *                      communicate with other MASes and        *
  *                      CMASes, an ability to be informed about *
  *                      non-responsive MASes is required.       *
  ****************************************************************
  * RECOMMENDATION: After applying the PTF that resolves this    *
  *                 APAR, all CMASes MASes must be restarted.    *
  *                 Note that the restarts do not need to occur  *
  *                 at the same time.                            *
  ****************************************************************
  When a MAS becomes non-responsive, it consumes resources that
  would normally be reused for communications with other MASes and
  CMASes.  If a MAS is non-responsive for a long period of time,
  it may consume all resources needed for communications, causing
  the CMAS to lose connectivity to all other MASes and CMASes.
  
  Since it is not possible to timeout the connection between a
  CMAS and MAS, it is necessary to have a warning system to alert
  users of non-responsive MASes, so that appropriate actions can
  be taken.
  
  To that end, method EYU0CSLT (CSLT) will run in a CMAS every 60
  seconds, polling all of the MASes to which the CMAS is
  connected, to determine if they are non-responsive.  CSLT will
  base its determination on two factors:
  
  1.  Is the MAS receiving requests sent from the CMAS?
  
  2.  Is the MAS returning required responses to the CMAS?
  
  If a MAS is determined to be non-responsive, new message
  EYUCS0207W will be issued by the connected CMAS.  If a MAS that
  had been determined to be non-responsive becomes responsive, new
  message EYUCS0208I will be issued by the connected CMAS.  The
  text of these new messages is as follows:
  
    EYUCS0207W  Connected MAS <masname> is not responding -
                APPLID(<masappl>) CICSplex(<plexname>)
  
    EYUCS0208I  Connected MAS <masname> is now responding -
                APPLID(<masappl>) CICSplex(<plexname>)
  

Problem conclusion

• To effect this change, the following updates have been made:
  
  -  the CMAS to MAS connection link conversation block (EYURCLCB)
     has been updated to include a flag to indicate whether a MAS
     is considered non-responsive, and fields used to determine if
     a MAS is non-responsive.
  
  -  copybook EYUUCMEQ and module EYUTCMM0 have been updated to
     define the new messages EYUCS0207W and EYUCS0208I.
  
  -  method EYU0CLIN (CLIN), which is called in a MAS when the
     connection to a CMAS is initialized, has been updated to set
     the MAS's APPLID into the EYURCLCB for the connection so it
     can be included in the new messages.
  
  -  method EYU0CLFX (CLFX), which is called in a CMAS when it
     adds a request to the MAS's send queue, and in the MAS when
     it retrieves a request from its send queue, has been updated
     to set the new field in the EYURCLCB that relates to whether
     the MAS is receiving requests sent from the CMAS.
  
  -  method EYU0CTBT (CTBT), which is called when a CMAS sends a
     request to a MAS, and in a MAS when it responds back to the
     CMAS with the results of the request, has been updated to set
     the new fields in the EYURCLCB that relate to whether the MAS
     is returning responses to the CMAS.  When called in the CMAS,
     it notes that an outstanding response is required.  When
     called in a MAS, it notes that a response has been returned.
  
  -  method EYU0CSLT (CSLT), which is the connection services long
     running task in the CMAS, has been updated to set a timer
     such that it is driven every 60 seconds to poll all connected
     MASes to determine if they are non-responsive, using the new
     EYURCLCB fields set by CLFX and CTBT.
  
     If a MAS is determined to be non-responsive during a polling
     interval and the EYURCLCB non-resonsive flag is not already
     turned on, the flag is turned on and message EYUCS0207W is
     issued.  Additionally, CSLT will issue an exception trace
     with a debug text of "NR_MAS" and a trace point Id of 39.
  
     If a MAS is determined to be responsive during a polling
     interval and the EYURCLCB non-responsive flag is turned on,
     the flag is turned off and message EYUCS0207I is issued.
  
  -  method EYU0CLET (CLET), which is called in a CMAS when the
     connection to a MAS is terminated, has been updated to turn
     off the non-responsive flag and clear the MAS's APPLID from
     the EYURCLCB for the connection.
  

Temporary fix

• FIX AVAILABLE BY PTF ONLY
  

Comments

• Since a single non-responsive MAS can
  result in a CMAS not being able to
  communicate with other MASes and
  CMASes, an ability to be informed about
  non-responsive MASes is required.
  
  
  When a MAS becomes non-responsive, it consumes resources that
  would normally be reused for communications with other MASes and
  CMASes.  If a MAS is non-responsive for a long period of time,
  it may consume all resources needed for communications, causing
  the CMAS to lose connectivity to all other MASes and CMASes.
  
  Since it is not possible to timeout the connection between a
  CMAS and MAS, it is necessary to have a warning system to alert
  users of non-responsive MASes, so that appropriate actions can
  be taken.
  
  To that end, method EYU0CSLT (CSLT) will run in a CMAS every 60
  seconds, polling all of the MASes to which the CMAS is
  connected, to determine if they are non-responsive.  CSLT will
  base its determination on two factors:
  
  1.  Is the MAS receiving requests sent from the CMAS?
  
  2.  Is the MAS returning required responses to the CMAS?
  
  If a MAS is determined to be non-responsive, new message
  EYUCS0207W will be issued by the connected CMAS.  If a MAS that
  had been determined to be non-responsive becomes responsive, new
  message EYUCS0208I will be issued by the connected CMAS.  The
  text of these new messages is as follows:
  
    EYUCS0207W  Connected MAS <masname> is not responding -
                APPLID(<masappl>) CICSplex(<plexname>)
  
    EYUCS0208I  Connected MAS <masname> is now responding -
                APPLID(<masappl>) CICSplex(<plexname>)
  
  
  To effect this change, the following updates have been made:
  
  -  the CMAS to MAS connection link conversation block (EYURCLCB)
     has been updated to include a flag to indicate whether a MAS
     is considered non-responsive, and fields used to determine if
     a MAS is non-responsive.
  
  -  copybook EYUUCMEQ and module EYUTCMM0 have been updated to
     define the new messages EYUCS0207W and EYUCS0208I.
  
  -  method EYU0CLIN (CLIN), which is called in a MAS when the
     connection to a CMAS is initialized, has been updated to set
     the MAS's APPLID into the EYURCLCB for the connection so it
     can be included in the new messages.
  
  -  method EYU0CLFX (CLFX), which is called in a CMAS when it
     adds a request to the MAS's send queue, and in the MAS when
     it retrieves a request from its send queue, has been updated
     to set the new field in the EYURCLCB that relates to whether
     the MAS is receiving requests sent from the CMAS.
  
  -  method EYU0CTBT (CTBT), which is called when a CMAS sends a
     request to a MAS, and in a MAS when it responds back to the
     CMAS with the results of the request, has been updated to set
     the new fields in the EYURCLCB that relate to whether the MAS
     is returning responses to the CMAS.  When called in the CMAS,
     it notes that an outstanding response is required.  When
     called in a MAS, it notes that a response has been returned.
  
  -  method EYU0CSLT (CSLT), which is the connection services long
     running task in the CMAS, has been updated to set a timer
     such that it is driven every 60 seconds to poll all connected
     MASes to determine if they are non-responsive, using the new
     EYURCLCB fields set by CLFX and CTBT.
  
     If a MAS is determined to be non-responsive during a polling
     interval and the EYURCLCB non-resonsive flag is not already
     turned on, the flag is turned on and message EYUCS0207W is
     be issued.  Additionally, CSLT will issue an exception trace
     with a debug text of "NR_MAS" and a trace point Id of 39.
  
     If a MAS is determined to be responsive during a polling
     interval and the EYURCLCB non-responsive flag is turned on,
     the flag is turned off and message EYUCS0207I is issued.
  
  -  method EYU0CLET (CLET), which is called in a CMAS when the
     connection to a MAS is terminated, has been updated to turn
     off the non-responsive flag and clear the MAS's APPLID from
     the EYURCLCB for the connection.
  

APAR Information

• APAR is sysrouted FROM one or more of the following:

• APAR is sysrouted TO one or more of the following:

  PK65669 UK37789

Modules/Macros

•    EYURCLCB EYUTCMM0 EYUTXLPD EYUUCMEQ EYU0CLET
  EYU0CLFX EYU0CLIN EYU0CSLT EYU0CTBT
  

Fix information

• Fixed component name

  CPSM CICS 3.1

• Fixed component ID

  5655M1501

Applicable component levels

• R100 PSY UK37789

     UP08/07/04 P F807

Fix is available

• Select the PTF appropriate for your component level. You will be required to sign in. Distribution on physical media is not available in all countries.