Dumps

Under exceptional circumstances, Windows MiniDumps, UNIX core dumps, or z/OS SVC or core dumps might be produced; for example, if errors occur in the broker or User Name Server executable files or in the infrastructure main program. The amount and complexity of data produced in these cases normally requires the assistance of your IBM Support Center.

Dumps on Windows

Windows MiniDumps might be produced by broker processes in extreme cases. They are never produced during normal operation.

They are accompanied by a BIP2111 error message (message broker internal error) that contains the path to the MiniDump file in your workpath/errors directory. MiniDump files have the extension .dmp. You can analyze these MiniDumps using a suitable debugger; contact IBM for assistance.

Dumps on UNIX systems

UNIX core dumps are produced by broker processes in extreme cases. They are never produced during normal operation.

A BIP2060 error message (execution group terminated unexpectedly) might be produced. Look in the directory where the broker was started to find the core dump file. If this directory is not writable by the service ID, the core dumps are produced in the service user ID's home directory.

You can analyze these core dumps using a suitable debugger; contact IBM for assistance.

Dumps on z/OS

A WebSphere Message Broker for z/OS broker or User Name Server produces different types of dumps depending on where the original error occurs.
SVC dumps
These are produced for errors in the WebSphere Message Broker for z/OS infrastructure main program (bipimain).
Core dumps
These are produced for errors in the broker or User Name Server executable files (bipservice, bipbroker, DataFlowEngine, and bipuns).

SVC dumps

Errors in the infrastructure main program cause system abend X'2C3' followed by a specific reason code. This abend might occur before the WebSphere Message Broker for z/OS recovery routine is active, so might occur without a dump being written. The reason code helps the IBM Support Center to determine the error. In that situation you get only the following information in the z/OS syslog:
IEF450I STU1051 - ABEND=S2C3 U0000 REASON=01050001

An SVC dump that is caused by z/OS or WebSphere Message Broker for z/OS is written to the system's dump directory, and can be formatted with IPCS. The name of the dump is listed on the z/OS syslog.

Under some circumstances, SVC dumps are not produced. Generally, dumps are suppressed because of time or space problems, or security violations. In addition, you can suppress SVC dumps that duplicate previous dumps using z/OS DAE, for example, use the command SET DAE=xx. (DAE is Dump Analysis and Elimination: a z/OS service that enables an installation to suppress SVC dumps and ABEND SYSDUMP dumps that are not needed because they duplicate previously written dumps.)

The following example shows a message on the z/OS syslog, indicating whether duplicates of SYSMDUMPs are suppressed:
   IEA995I SYMPTOM DUMP OUTPUT 504
 SYSTEM COMPLETION CODE=0C4  REASON CODE=00000004
  TIME=11.02.24  SEQ=02327  CPU=0000  ASID=0060
  PSW AT TIME OF ERROR  078D2000   8D70A656  ILC 4  INTC 04
    ACTIVE LOAD MODULE           ADDRESS=0D708F68  OFFSET=000016EE
    NAME=SPECIALNAME
         61939683 81936199 85A2F1F0 61A48689 */local/res10/ufi*
         A7F5F161 82899561 82899789 94818995 *x51/bin/bipimain*
    DATA AT PSW  0D70A650 - 91345000  00001F22  41209140
    GPR  0-3  00000000  0D711B18  0D711B18  00000000
    GPR  4-7  0D711300  0D70BD04  00000000  0D711B18
    GPR  8-11 0D7121CF  0D7111D0  0D70C316  0D70B317
    GPR 12-15 8D70A318  0D7111D0  00000312  00000000
  END OF SYMPTOM DUMP
 IEA838I SYSMDUMP SUPPRESSED AS A DUPLICATE OF: 505
         ORIGINAL:DATE 01170 TIME 10:59:40:05 CPU 8A7645349672
         MOD/#PATHNAM CSECT/BIPIP PIDS/5655G9700 AB/S00C4
         REXN/IMBSIREC FI/9134500000001F2241209140 REGS/0C33E
         HRC1/00000004 SUB1/INFRASTRUCTURE#MAIN           

The MVS Diagnosis: Tools and Service Aids manual provides details about using z/OS dump analysis and elimination (DAE).

Core dumps

A broker or User Name Server produces USS (UNIX System Services) core dumps. These core dumps are SYSMDUMP dumps. They are written to the started task's user directory. The name is coredump.pid, where .pid is the hexadecimal value of the process ID of the process that encountered the error, and is listed on the z/OS syslog. The dump can be found in the started task users' home directory, for example, /u/stu1/coredump.000000ae.

The maximum size of a core dump is defined through MAXCORESIZE in the BPXPRMxx parmlib member. The IBM-supplied default is 4MB. To ensure completeness of a core dump of any WebSphere Message Broker for z/OS address space, change this to 2GB. The started task user's directory must then have at least this size.

When the error recovery routines of WebSphere Message Broker for z/OS provide an SVC dump, a core dump is also written to your started task user ID directory. In this case, the SVC dump and the entry on the z/OS syslog are of interest to IBM to resolve the problem. The following example shows an entry on the z/OS syslog when an SVC dump and core dump have been provided:
IEA794I SVC DUMP HAS CAPTURED: 486                                      
 DUMPID=008 REQUESTED BY JOB (STU1051 )                                 
 DUMP TITLE=COMP=WMQI z/OS CID=G9700 ISSUER=BIPIMAIN (BIPIP             
            ): WMQI InfraMain         : S0C4-00000004.                  
 IEF196I IGD101I SMS ALLOCATED TO DDNAME (SYS00010)                     
 IEF196I         DSN (SYS3.DUMP.ARG5.STU1051.T085940.S00008       )     
 IEF196I         STORCLAS (SMS) MGMTCLAS (DUMP) DATACLAS (        )     
 IEF196I         VOL SER NOS= ARGSMY                                    
 IEF196I IGD104I SYS3.DUMP.ARG5.STU1051.T085940.S00008        RETAINED, 
 IEF196I DDNAME=SYS00010                                                
 IEA611I COMPLETE DUMP ON SYS3.DUMP.ARG5.STU1051.T085940.S00008 493     
 DUMPID=008 REQUESTED BY JOB (STU1051 )                                 
 FOR ASID (0060)                                                        
 INCIDENT TOKEN: ARGPLEX  ARG5     06/19/2001 08:59:40                  
  ERROR ID = SEQ02325 CPU00 ASID0060 TIME10.59.40.5                     
 IEA995I SYMPTOM DUMP OUTPUT 494                                        
 SYSTEM COMPLETION CODE=0C4  REASON CODE=00000004                       
  TIME=10.59.40  SEQ=02325  CPU=0000  ASID=0060                         
  PSW AT TIME OF ERROR  078D2000   8D70A656  ILC 4  INTC 04             
    ACTIVE LOAD MODULE           ADDRESS=0D708F68  OFFSET=000016EE      
    NAME=SPECIALNAME                                                  
         61939683 81936199 85A2F1F0 61A48689 */local/res10/ufi*       
         A7F5F161 82899561 82899789 94818995 *x51/bin/bipimain*       
    DATA AT PSW  0D70A650 - 91345000  00001F22  41209140              
    GPR  0-3  00000000  0D711B18  0D711B18  00000000                  
    GPR  4-7  0D711300  0D70BD04  00000000  0D711B18                  
    GPR  8-11 0D7121CF  0D7111D0  0D70C316  0D70B317                  
    GPR 12-15 8D70A318  0D7111D0  00000312  00000000                  
  END OF SYMPTOM DUMP                                                 
 IEA993I SYSMDUMP TAKEN TO coredump.0300019D
To make use of these core dumps, copy them to a partitioned data set using the TSO/E OGET command. Allocate the data set as a sequential data set with a logical record length (LRECL) of 4160 and a record format of FBS. Specify a primary allocation of at least 500 cylinders and a secondary allocation of at least 250 cylinders. The data set must be transferred as binary. For example:
oget '/u/user_directory/coredump.pid' 'mvs_dataset_name.pid' bin

Dump options

To collect all the diagnostic information, specify the following dump options in SYS1.PARMLIB:
  • Member IEADMP* SDATA=(LSQA,TRT,CB,ENQ,DM,IO,ERR,SUM)
  • Member IEADMR* SDATA=(NUC,SQA,LSQA,SWA,TRT,RGN,LPA,CSA,SUM,GRSQ)
You can check the values on your system by using the D Dump,Options z/OS console command, as follows:
D D,O
Example output from this command is as follows:
    SYSABEND- ADD PARMLIB OPTIONS SDATA=(LSQA,TRT,CB,ENQ,DM,IO,ERR,SUM),
                         PDATA=(SA,REGS,LPA,JPA,PSW,SPLS)
    SYSUDUMP- ADD PARMLIB OPTIONS SDATA=(LSQA,TRT,CB,ENQ,DM,IO,ERR,SUM),
                         PDATA=(SA,REGS,LPA,JPA,PSW,SPLS)
    SYSMDUMP- ADD OPTIONS (NUC,SQA,LSQA,SWA,TRT,RGN,LPA,CSA,SUM,GRSQ)
The last entry shows the SYSMDUMP options that are used when taking core dumps.
Related reference
Local error logs
Abend files