If IMS dependent regions cannot issue SQL statements
This is a situation in which users communicating to Db2 through IMS cannot complete transactions, but Db2 is still operational, still responding to commands, or operating normally through an attachment facility.
- Issue the IMS /DISPLAY ACTIVE command repetitively to identify the transactions being processed by each dependent region. Any dependent regions whose transaction codes are not changing might be involved in a WAIT or LOOP.
- Issue the IMS /DISPLAY TRAN xx to verify that the program/transactions are not stopped.
- Issue the z/OS® command DISPLAY ACTIVE jobname repetitively, once for each message region that appears to be in a WAIT or LOOP.
- Determine if the processor time remains the same. If so, the dependent region is probably in a WAIT state. Follow the steps in Gathering the information.
- Otherwise, determine if the dependent region is in a loop.
- IMS detects loops that occur within IMS dependent regions. The user specifies the timeout value using the TRANSACTION macro PROCLIM parameter.
- If the dependent region has a long timeout, issue the IMS /DISPLAY TRAN y command. If this shows the IMS transaction is decreasing, the region is processing messages and the application is not in an endless loop.
- If you want you can issue the Db2 command
-DISPLAY THREAD and review the
REQ
column, which often indicates loops.
- Issue the IMS /DISPLAY SUBSYSTEM command to determine the status of the connection to Db2.
- Issue the IMS /TRACE SET ON SUBS OPTION LOG command to trace IMS calls to Db2 on the IMS log.
- Issue the following IMS commands
in order until one breaks the connection. If successful, skip the
next step. If not successful, go to the next step.
- /STOP region reg# abdump
- /STOP region reg# CANCEL
- /STOP SUBSYS subsysname
- Issue the following MODIFY commands in this order until one successfully
breaks the subsystem connection. Do not confuse these with the /MODIFY
command or the MODIFY utility.
- F IMS,DUMPname
- F IMS,STOPname
- F IMS,FORCEname
name refers to the attached external subsystem name defined in the IMS external subsystem PROCLIB member.
Gathering the information
- If the reason for the wait or loop has not been determined, issue the IMS /STOP REGION x ABDUMP yyyy command to terminate the specific region involved in the wait or loop. If you receive a dump, the problem is in the application's address space and is probably the result of a user error. Analyze the dump and correct the problem. Do not continue with this procedure.
- If the z/OS GTF trace facility is not active, issue the z/OS START GTF command with the USR and TIME=YES options.
- Verify that dump data sets are available.
- Use the z/OS DUMP command to dump the Db2 address spaces (database services, system services, the distributed data facility address space for distributed processing only, and any user address spaces) and the IRLM address space. This generates many very large dumps, which can help to diagnose the problem if other methods fail.
- Terminate the dependent region and produce a SYSABEND or SYSUDUMP. Be aware that doing so can cause the IMS control to ABEND, with an ABENDU113, or cause Db2 to ABEND. With a dump data set available, issue the IMS /STOP REGION x CANCEL command.
- Print the entries in the SYS1.LOGREC data set for the estimated time that the loop or wait began. For more information on printing SYS1.LOGREC entries, refer to SYS1.LOGREC.
- Analyze the TOD (time of day) information in SYS1.LOGREC to determine if any failures occurred around the time the wait or loop occurred. Find entries related to the Db2 or dependent region address spaces.
- Determine if users are doing something in the IMS attachment facility that might prevent them from accessing Db2. This might indicate a wait or loop in another component or in the attachment facility.
- Determine if control is in IMS or in Db2. Review the information in the SYSABEND dump, IMS attachment facility trace table, and Db2 trace table. The trace tables are described in Printing and analyzing global traces.
- If control is in IMS, refer to IMS diagnosis information. Otherwise, keep the SYS1.LOGREC entries, the GTF trace, and the dumps you obtained in Step 4.
- Add
WAITorLOOPto the keyword string. If unsure about which is appropriate, create two strings, one with each keyword. Turn to Searching the IBM Support site for known problems and solutions.(free format): 5740XYR00 R111 WAIT 5740XYR00 R111 LOOP (structured format): PIDS/5740XYR00 LVLS/111 WAIT PIDS/5740XYR00 LVLS/111 LOOP