Getting dump of TSO user

Answer

When a user doing a "*" type search against the CP causes other users to go into a "wait" it is possible to get a dump of the suspected address space.

Make certain the system dump parms include all the storage areas listed in the TWZSz Installation Guide under the topic " UPDATING MVS DUMP OPTIONS" in Chapter 4 - INSTALLING.

The following are the steps to follow in order to get a "dump" when a 'hang' occurs:

1. To get a DUMP when a 'hang' problem occurs, issue the "DUMP" command for the suspected TSO user's logon address space before issuing the "cancel" command. In other words, from another TSO session, while the "hang" is in progress, issue the following command;

"DUMP COM=(TWS DIALOG HANG)"

2. The system will respond with message:

"IEE094D SPECIFY OPERAND(S) FOR DUMP COMMAND"

and will prompt for the ASID.

3. Respond with "ASID=xxxx" where "xxxx" is the HEX ASID of the suspect address space.

NOTE: Do not cancel the failing user's session until after the dump has completed.

4. There is also an "EQQDUMP" dataset that should be allocated to ALL TWS DIALOG users. It is really more like a trace file than a dump. "EQQDUMP" must be allocated in your "TSO LOGON PROC" or a "CLIST" that is called from the TSO LOGON proc.

NOTE: See sections for "The Diagnostic File" (EQQDUMP) and "Information needed for all problems" in Chapter 3 of the TWSz Diagnosis Guide and Reference.

5. When attempting to capture an actual "SYSMDUMP" of the hang, dump the OPC Controller address space and the suspect TSO USER address space while the "hang" is occurring . CANCEL dumps are not useful for problem determination.

6. Issue a "D GRS,ALL" command on the MVS console during the "hang" to show if there is any "ENQ" contention.

The full set of documenation is the following::
(1) EQQDUMP output from the TSO user "causing" the hang
(2) DUMP of TWS controller and TSO user address spaces
(3) Output of the D GRS,ALL command
(4) TWS controller MLOG

SOLUTION # 114730

.