OA22693: CSM AND TIPAC STORAGE GROWTH. XCF CONNECTION HUNG DUE TO MISMANAGED XCF PACING RESPONSE.

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• HPR sessions using XCF failed with sense 800D0000 for AM RESUME
  request. CSM and TIPAC (TIBUF) storage is growing.  Work
  elements are queued to the NCBWORKQ (MPNCB) for an XCF
  connection, but are not being processed.  The XCF connection
  is hung due to a mismanaged XCF pacing response.
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED: All using XCF connections.                   *
  ****************************************************************
  * PROBLEM DESCRIPTION: XCF connection hangs indefinitely       *
  *                      due to mismanaged XCF pacing response.  *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  The problem is summarized as follows:
  1) An XCF connection is being utilized for SNA and TCP/IP
     traffic.
  2) To avoid overrunning the partner XCF node, the XCF
     connection is paced. The current pacing window is 500
     messages.
  3) At some point, one end of the XCF connection suspended
     outbound data processing, waiting for the arrival
     of an XCF pacing response.
  4) When this occurs, the XFNCB_WAITING_FOR_PACRSP bit is
     turned on by module ISTTSC8W.  Processing in
     module ISTTSC8W seems to suspend for a brief moment.
  5) At the same instant, the XCF pacing response is
     received by the XCF message exit.
  6) When the pacing response is received, the
     XFNCB_WAITING_FOR_PACRSP bit is turned off.  The
     SCL_STATE is then set to SCLIDLE by ISTTSCBX/ISTTSCDX.
     The MPNCB PC PAB is then scheduled to resume outbound
     work.
  7) In this case, module ISTTSC8W resumes the "wait for
     pacing" processing by setting the SCL_STATE to
     SCLBUSY.  The current TOD clock is also stored in the
     XFNCB_Wait4Pac_TOD field.
  8) When the MPNCB PC PAB dispatches (ISTTSC8S), it
     cannot process the outbound work as the SCL_STATE is
     SCLBUSY (and not SCLIDLE as expected).
  9) This synchonization problem leads to the XCF connection
     hanging.
  10) Over time, TCP/IP and SNA traffic builds up on the outbound
      XCF MPNCB NCBWORKQ leading to storage exhaustion.
      Specifically, CSM and TIPAC buffer pools were growing
      over time.
  

Problem conclusion

• ISTTSC8W - Has been changed synchronize the
             setting of the SCL_STATE and
             the XFNCB_Wait4Pac_TOD fields with
             modules ISTTSCBX and ISTTSCDX.  This is
             done by setting these fields before
             updating the XFNCB_SWAP fields (CS logic).
  ISTTSCMA - Included for maintenance purposes.
  

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:

  UA37370 UA37371 UA37372

Modules/Macros

• ISTTSCMA ISTTSC8W
  

Fix information

• Fixed component name

  VTAM V4 MVS/ESA

• Fixed component ID

  569511701

Applicable component levels

• R170 PSY UA37370

     UP07/11/14 P F711 Ž

• R180 PSY UA37371

     UP07/11/14 P F711 Ž

• R190 PSY UA37372

     UP07/11/07 P F711 Ž

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.