PH07028: z/OS Connect EE API requester calls are repeated after a timeout failure.

APAR status

• Closed as unreproducible in next release.

Error description

• A z/OS  API requester application ran as a batch job and made
  multiple outbound API requests sequentially to a z/OS Connect EE
  server. In a test with 6 requests, it was found that the z/OS
  Connect EE communications stub for IMS and z/OS applications
  sent 8 requests, the 6 intended requests and 2 duplications.
  
  The z/OS Connect EE communication stub timeout (BAQTIMEOUT) was
  set to 5 seconds but the target API providers sometimes take
  more than 5 seconds to respond. The communication stub trace
  showed 3 successful requests with http status 200 and 5 requests
  returned error code 1102:
  
  t: recv() failed.
  errno: 1102 error: EDC8102I Operation would block.
  t: Unable to read data from socket.
  t-Entry: error
  t: An error occurred: EDC8102I Operation would block.
  t: Reason code: 1102
  t: Return code: -1
  t: Service: 9
  t: Service Instance: 0
  t-Exit: error
  t-Entry: setReturnCode
  t-Exit: recvresp
  t: Send request returned with a non-reconnect error code: 1102
  t-Exit: sendrqst
  
  There were two instances in the trace to try reconnect, despite
  it was a 'non-reconnect error code: 1102' The retry request
  failed again with error 1102, then the requests terminated.
  
  Additional search words: BAQCSTUB msgEDC8102I msgEDC8102 EDC8102
  rc1102
  

Local fix

• Set a shorter timeout value between z/OS Connect EE and the API
  endpoint than the value for the timeout between the API
  requester application and the z/OS Connect EE server.
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: All users of z/OS Connect EE V3 API          *
  *                 requester for IMS and z/OS applications.     *
  ****************************************************************
  When a z/OS Connect EE API requester request from an IMS or
  z/OS application failed due to the BAQTIMEOUT being exceeded,
  the z/OS Client Web Enablement Toolkit returned a non-retryable
  error code 1102. The communications stub erroneously retries the
  request leading to duplicate requests.
  

Problem conclusion

Temporary fix

Comments

• z/OS Connect EE has been changed to improve the handling of
  communication failures and timeouts of API Requester calls for
  IMS and z/OS applications. When a communication error is
  received, BAQCSTUB will close the z/OS Client Web Enablement
  Toolkit socket connection. A new connection will be established
  when the next request is received. This prevents the reported
  problem of requests being duplicated after a timeout.
  
  The fix for this APAR is expected to be delivered by the PTF for
  z/OS Connect EE V3.0.21.0 (PH11043).
  

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

  Z/OS CONNECT EE

• Fixed component ID

  5655CE300

Applicable component levels