David,
IIRC there were several discussion of CATALOG CPU use in the last year. The
suggestion in several cases was to start by getting current on service. Why
not apply the latest CATALOG level set PTFs and then if this doesn't improve
open a PMR with IBM to discuss.
It sounds like understanding
In [EMAIL PROTECTED], on 06/10/2005
at 04:31 PM, Edward E. Jaffe [EMAIL PROTECTED] said:
Yes. That's what you wrote. But Chris wrote it a day earlier.
Whatever he may have written in a different message, his response to
my message was not a day earlier than my message ;-)
--
Shmuel
In
[EMAIL PROTECTED],
on 06/10/2005
at 06:46 PM, Craddock, Chris [EMAIL PROTECTED] said:
1. that work done by a TCB is charged to the TCB and its home address
space (or enclave) EVEN WHEN it runs in cross memory mode in another
address space. I count cross memory time as time spent in another
1. that work done by a TCB is charged to the TCB and its home address
space (or enclave) EVEN WHEN it runs in cross memory mode in another
address space. I count cross memory time as time spent in another
address space.
I guess that we will have to agree to disagree as to whether it is the
Ed said
Such function (CAS WLM exploitation) takes time to
design and implement. At present, it's likely near the bottom of the
needs to get done in a hurry queue.
Especially since Mark is off draining the re-work VSAM swamp.
I'd guess almost all functional changes in CATALOG would be on
In [EMAIL PROTECTED],
on 06/09/2005
at 06:29 PM, Craddock, Chris [EMAIL PROTECTED] said:
All of the work that runs under the requestor's TCB is charged to the
requestor's TCB, but the work that is done in CAS by the CAS TCB is
charged to CAS
Isn't that what I wrote?
--
Shmuel
In [EMAIL PROTECTED], on 06/09/2005
at 07:24 PM, Bill Fairchild [EMAIL PROTECTED] said:
I am confused.
Be careful to distinguish between dispatching status and addressing
modes.
Ed Jaffe says when in Xmem mode the home address space is charged
for CPU time (he meant the home's current
Shmuel said
, Craddock, Chris [EMAIL PROTECTED] said:
All of the work that runs under the requestor's TCB is charged to the
requestor's TCB, but the work that is done in CAS by the CAS TCB is
charged to CAS
Isn't that what I wrote?
It is probably what you meant, but not quite what you
I still think my Douglas Adams comment was as good a piece of advice as
anything I've encountered in this thread. To wit: I don't recall anyone
pointing out that JES2 EXIT 6, the one at issue in the original question,
runs in the USER environment, not the JES2 environment. That is, JES2 gives
Sounds like a capture ratio problem.
-Original Message-
From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On
Behalf Of Ed Finnell
Sent: Friday, June 10, 2005 9:47 PM
To: IBM-MAIN@BAMA.UA.EDU
Subject: Re: Can we measure cpu use of a JES2 Exit
In a message dated 6/10/2005 7
[EMAIL PROTECTED] wrote:
I must be missing something here, but here goes.
We used to test all of our JES2 exits by writing simple driver programs.
Just run the driver program with and without the Exit call and subtract
the differnce.
Subtract *what* difference? How are you measuring all
In a message dated 6/9/2005 10:09:35 A.M. Central Daylight Time,
[EMAIL PROTECTED] writes:
Subtract *what* difference? How are you measuring all the CPU time spent
in the CATALOG address space in response to SVC 26 requests from the
exit? (Haven't we been here already?)
Doesn't the
In a message dated 6/9/2005 10:37:03 A.M. Central Daylight Time,
[EMAIL PROTECTED] writes:
If that were the case for SVC 26 in general, would you expect CATALOG to
be such a large CPU consumer?
Right. I wouldn't. I don't know how it works. I'm asking about CPU timing
in general. If
Bill Fairchild wrote:
Right. I wouldn't. I don't know how it works. I'm asking about CPU timing
in general. If your primary address space PCs into a secondary, does the
secondary's CPU time get charged to your primary's TCB?
CPU time is charged to the home address space.
--
In [EMAIL PROTECTED], on 06/09/2005
at 11:14 AM, Bill Fairchild [EMAIL PROTECTED] said:
Doesn't the Catalog Address Space get involved because of a PC from
JES2?
Yes.
Wouldn't the TCB in the primary address space get charged for the
time?
No. The issuing TCB would be charged for time
Bill Fairchild wrote:
OK. Now I get it. I thought all the work was done synchronously under the
original SVC. Maybe this design makes it easier for WLM, but it would seem
to make it more difficult for chargeback.
But, as Chris correctly points out, this could be remedied by
We have JES2 exit 6 active on two lpars and have the approval to disable
it. This exit, among other things, does a number of svc 26 lookups. We
would like to be able to quantify how much cpu utilization is reduced by
turning off the exit. Can this be measured?
Thanks again.
Alan Schwartz
Alan,
A product called Stobe from Compuware may accomplish this.
Kevin.
-- Original message --
We have JES2 exit 6 active on two lpars and have the approval to disable
it. This exit, among other things, does a number of svc 26 lookups. We
would like to be
In a message dated 6/8/2005 1:57:29 P.M. Central Daylight Time,
[EMAIL PROTECTED] writes:
We have JES2 exit 6 active on two lpars and have the approval to disable
it. This exit, among other things, does a number of svc 26 lookups. We
would like to be able to quantify how much cpu
Alan Schwartz wrote:
We have JES2 exit 6 active on two lpars and have the approval to disable
it. This exit, among other things, does a number of svc 26 lookups. We
would like to be able to quantify how much cpu utilization is reduced by
turning off the exit. Can this be measured?
If
On Wed, Jun 08, 2005 at 12:30:26PM -0700, Edward E. Jaffe ([EMAIL PROTECTED])
wrote:
If you want to know how much CPU time an exit uses, don't remove it! Put
a TIMEUSED macro at entry and exit --
[snip]
What about JES2's overhead in invoking the exit? In the particular
case in question
On Wed, 2005-06-08 at 12:30 -0700, Edward E. Jaffe wrote:
If you want to know how much CPU time an exit uses, don't remove it! Put
a TIMEUSED macro at entry and exit
A good suggestion, Ed -- but will it capture time spent in the CATALOG
address space (as a side effect of those SVC 26 calls)?
David Andrews wrote:
On Wed, 2005-06-08 at 12:30 -0700, Edward E. Jaffe wrote:
If you want to know how much CPU time an exit uses, don't remove it! Put
a TIMEUSED macro at entry and exit
A good suggestion, Ed -- but will it capture time spent in the CATALOG
address space (as a side
Has anyone checked to see if the $DPERFDATA(*) (I think that is it) command
would help? I don't have access to JES2 at the moment, but do remember
using it to get a vast amount of internal statistical data out of JES2 with
it.
Lizette Koehler
On Wed, 8 Jun 2005 13:57:21 -0500, Alan Schwartz
[EMAIL PROTECTED] wrote:
We have JES2 exit 6 active on two lpars and have the approval to disable
it. This exit, among other things, does a number of svc 26 lookups. We
would like to be able to quantify how much cpu utilization is reduced by
25 matches
Mail list logo
