Phil,
>From messages manual for IEF374I which formatted 46.98 seconds for your
first case
CPU xxxx MIN xx.xx SEC
For processor time, which includes enclave time, preemptive class
SRB
time, client SRB time, and normalized IFA service time, xxxx MIN
specifies the minute and xx.xx SEC specifies the second (in seconds
and hundredths of a second).
And from z/OS V1R9.0 MVS Installation Exits SA22-7593-12 for IEFACTRT
Word 6
Address of a 4-byte area. The first 3 bytes contain the step processor
time (which includes enclave time, preemptable class SRB time, and
client SRB time) under TCBs, in hundredths of a second (in binary). The
last byte contains the number of accounting fields in the EXEC statement
(in binary). At job termination, the field is zero.
Note:
Since this step processor time is only 3 bytes it is limited to 46
hours. If the step accumulates more than this amount of processor time,
this field will become invalid. You can avoid this situation by having
your exit get the processor time from the parameter at word 12.
The SYS1.SAMPLIB(IEEACTRT) formats word 6 in the TCB time field (200.44)
in your first case. It isn't clear what fields are used to calculate
this value.
You need to examine the steps SMF type 30 record to understand the
various components of CPU time.
Is any of the work zAAP enabled (what IEF374I calls IFA) - JAVA or XML
parsing? Interestingly IEF374I says zAAP time is normalized but word 6
doesn't mention zAAP time.
If you have multiple tasks and multiple processors (including zAAP
processors) than CPU time can exceed clock time (.9 in first case).
If you have system command authority do a D M=CPU to determine number,
type and availability of processors. You will see something like this:
PROCESSOR STATUS
ID CPU
00 + *
01 + * 3 general processors
02 + *
03 +A <- zAAP processor
04 +I <- zIIP processor
Dave Kreiss
-----Original Message-----
From: IBM Mainframe Discussion List [mailto:[email protected]] On
Behalf Of Phil Smith III
Sent: Friday, May 22, 2009 10:36 AM
To: [email protected]
Subject: TCB time question
The following is typical output found in the SYSMSG produced by a batch
job on our z/OS 1.9 system:
-STEPNAME PROCSTEP RC EXCP CONN TCB SRB CLOCK SERV
-NNN1SW 00 2326 1195 200.44 .00 .9 3094K
-NNN1HW 00 2316 1069 .64 .00 .0 9753
The following are the step termination message produced for those steps:
/START 2009139.1224
/STOP 2009139.1225 CPU 0MIN 46.98SEC SRB 0MIN 00.03SEC
/START 2009139.1225
/STOP 2009139.1225 CPU 0MIN 00.15SEC SRB 0MIN 00.01SEC
We are not able to reconcile the step TCB time, particularly in the
first step, with the actual run/clock time of that step. In fact, the
TCB time attributed to the first step is considerably greater than the
job's entire wall-clock run time.
The ratios match -- 200.44/.64 ~= 46.98/.15 -- but what units is the
"200.44" in? What is the TCB time showing us?
Thanks,
.....phsiii
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
