OA36773: ECSA GROWTH IN 32K CSM BUFFERS WHEN USING HIPERSOCKETS DEVICES

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• During periods of very heavy HiperSockets traffic, storage
  growth can be seen in 32K ECSA CSM buffers.  The buffers will
  contain IUDR eyecatchers.  There will also be SKMB control
  blocks pointing to the IUDRs, but not chained on any connection
  control blocks.
  .
  VERIFICATION STEPS:
  1) Take a dump of TCPIP and VTAM
  2) The dump will show growth in subpool 231 key 6.  The storage
  will contain IUDR control blocks
  3) CSM tracker should be enabled to track the current owner of
  the storage:
  MODIFY VTAM,CSM,MONITOR=YES
  4) The CSM tracker will show the storage owned by ID x'20'
  .
  KEYWORDS:
  sp231k6 sp231 k6 CSA ECSA SKDB
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED: All using Hipersockets Multiwrite            *
  *                 with ZIIP Assist.                            *
  ****************************************************************
  * PROBLEM DESCRIPTION: Storage growth of 32K ECSA CSM buffers  *
  *                      on a HiperSockets Multiwrite device     *
  *                      with ZIIP enabled under periods of      *
  *                      heavy traffic.                          *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  While running Hipersockets, ISTLLCHI is called to process work
  passed from IUTLLCIZ on a ZIIP thread being passed by a zIIP
  enabled application (i.e. DB2).
  
  The data is written into the SBALs, but the ring becomes
  completely full before all the passed work can be processed.
  In that case, ISTLLCHI exits without requeuing the passed
  work, thereby leaving the data buffers stranded.  Over time,
  this can constrain CSM storage.
  

Problem conclusion

• ISTLLCHI has been changed to ensure leftover work passed
  directly from IUTLLCIZ gets requeued whenever the SBAL ring
  gets full or the maximum number of SBALs have been processed.
  
  ISTLLCWC have been changed handle Random Early Slowdown (RES)
  if the SBAL ring gets full.
  
  ISTLLCM8 has been included for maintenance purposes only.
  

Temporary fix

• *********
  * HIPER *
  *********
  

Comments

APAR Information

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

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

  UA61436 UA61437 UA61438 UA61439

Modules/Macros

• ISTLLCHI ISTLLCM8 ISTLLCWC
  

Fix information

• Fixed component name

  VTAM V4 MVS/ESA

• Fixed component ID

  569511701

Applicable component levels

• R1A0 PSY UA61436

     UP11/09/02 P F109

• R1B0 PSY UA61437

     UP11/09/02 P F109

• R1C0 PSY UA61438

     UP11/09/02 P F109

• R1D0 PSY UA61439

     UP11/09/02 P F109

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.