ABEND878 WebSphere MQ for z/OS due to DXWB control blocks filling storage

Technote (troubleshooting)


You are seeing ABEND878 in the WebSphere MQ MSTR address space and/or the CHIN address space.


Abends might be observed in the WebSphere MQ MSTR and or CHIN and, since storage is affected, other unrelated applications might also exhibit unexpected symptoms. In a queue-sharing group the WebSphere MQ MSTR private storage is filled with an unexpectedly high number of DXWB control blocks


An address space has been swapped out

Diagnosing the problem

A dump of the WebSphere MQ MSTR address space is used to analyze SP229 Key 7. This local storage contains XCF Group exit DXWBs and many are in a 'Waiting' state. The CHIN can terminate with a general error code of ABEND5C6-00E7014F, but the dump workarea contains an internal reason code indicating that a local storage shortage has occurred. CSQY220I in the WebSphere MQ MSTR joblog shows a rapid increase in the amount of storage used (because of DXWB control blocks). These blocks are used to queue requests from XCF to the WebSphere MQ user or message exits. If these requests can not be processed (due to some problem affecting the Coupling Facility) then their count grows, leading to storage constraint. Reviewing a dump of SRMDATA reveals the swapped out address space that might have been subject to a RESET QUIESCE (indicated by 'ADDRESS SPACE QUIESCED' )

JOB   12345678                                        
ASID  0010                                          
OUCB  12345678 OUT     QUEUE                          
                +10 (LSW)  LOGICALLY SWAPPED          
                +18 (RQSC) ADDRESS SPACE IS QUIESCED  
                +29 (SRC)  SWAP OUT REASON: UNILATERAL
                (ASCBRSME) RAX ADDRESS IS 2345678    
                           SERVICE CLASS = BATCH00    
                           WORKLOAD = BATCH          
                           PERIOD = 02              

Resolving the problem

While technote 1443264 documents other causes of DXWB build-up, in this case the address space is logically swapped out as a result of the operator RESET QUIESCE command. A quiesce only ends if (1) the job ends, or (2) a RESET RESUME command is issued. If the address space is swapped out in the midst of an MQ API call with latches held, then other resources requiring those latches can not attain them until the address space is swapped back in.

Within WebSphere MQ, in order to prevent batch jobs from being swapped out while performing MQ API calls, the relatively new CONNSWAP=NO parameter can be specified in CSQSYSP.

To avoid swapping in this scenario RESET QUIESCE should be used carefully.

Product Alias/Synonym

WebSphere MQ WMQ

Rate this page:

(0 users)Average rating

Document information

More support for:

WebSphere MQ
Storage Constraint (SOS)

Software version:

7.0, 7.0.1, 7.1

Operating system(s):

OS/390, z/OS

Reference #:


Modified date:


Translate my page

Machine Translation

Content navigation