Re: Another fine mess

Replies in [[ ]]

IBM Mainframe Discussion List <IBM-MAIN@xxxxxxxxxxx> wrote on 04/12/2006
02:11:13 PM:

Are you SURE that your batch job on LPAR 1 is the problem? That is, are
you
sure there is not some other factor outside of what you have described
here?
Is it possible something else is deleting or renaming C?

[[I own C. If at any point, there is no C, jobs begin to fail as the one
on lpar 2 did. At peak period, jobs can fail every minute. As for any
other issues, I am looking for suggestions.]]

The reason I ask is that it seems to me that even if you totally dropped
the
ENQs the situation you described would be unlikely. I don't mean you
should
drop the ENQs or that the situation would be sufficiently unlikely for
mainframe production, I just mean that you would be unlikely to hit it
in
any given day, and therefore this must be something else. File N does
not
exist for only a fraction of a second (rename to rename) - what's the
odds
that a batch job would hit that more than once in a blue moon? (Again,
too
often to go that way in production, but I am just doing a mental
exercise
here.)

[[The odds are very slim. And it happens every so often. It's not a big
hole I am closing. It is a hole. And I can't find the mouse. I just
need the name of the mouse so I know whose mouse it is.]]

After BS689 fails, does C still not exist? Or has it reappeared?

[[In human time the dataset always exists. Only computers can see the
hole in time the cybermouse goes through.]]

Any chance that the rename in your job is failing and you are not
checking
return codes correctly?

[[If rename fails, and there is no C, subsequent jobs fail. If rename
fails, and there is a C, I could detect that N still existed and that
dataset timestamps were wrong.]]

Have you checked timestamps? Have you confirmed that BS689 failed at the
exact instant that your job was doing the rename?

[[Timestamps the same to the second. Both steps take less than a second.

It should be mentioned that in the past, to correct this kind of problem,
rename was limited only to the cluster name of the KSDS since failure
would happen between the rename of the cluster and the rename of the
components.

This problem has been happening sporadically for several years.]]


Charles

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:IBM-MAIN@xxxxxxxxxxxx
Behalf
Of Kirk Talman
Sent: Wednesday, April 12, 2006 10:41 AM
To: IBM-MAIN@xxxxxxxxxxx
Subject: Another fine mess

Our tech people can't find the source of this problem and I am looking
for
any guesses you might have.

Sysplex in question has 11 lpars on 5 CECs

zOS 1.4 with 1.7 being tested. Sometimes this summer some of the lpars
in
this plex will be running 1.7 if all goes well. (So far it hasn't.)

I am told GRS used to propagate ENQ in star configuration.

Batch job on lpar 1 builds VSAM file called N. At the end of the build
job is a step using a program to frontend IDCAMS. The frontend issues
an
ENQ RET=NONE with scope of SYSTEMS. It then links to IDCAMS. IDCAMS
renames C to O and renames N to C, effectively bringing a new version of
C
into play. When IDCAMS ends, the frontend issues DEQ.

Meanwhile on lpar 2, a batch job using program called BT689 wants to use

file C. It calls a linked subroutine BS689 which issues an ENQ with the

same scope as mentioned above. The ENQ does not specify RET=NONE, but
that is the default. It does specify RNL=NO, which is not the default.
That should not matter since both ENQ have scope of SYSTEMS.

After it obtains the ENQ, BS689 does SVC 99 to dynamically allocate the
file. BT689 uses the file for a few reads and then ends.

Previously the job on lpar 1 would fail periodically because the file
was
in use.

This time the program BT689 on lpar 2 failed with message in its log

R15=>00000004 S99INFO=>00000002 S99ERROR=>00001708

IKJ56228I DATA SET C NOT IN CATALOG OR CATALOG CAN NOT BE ACCESSED



-----------------------------------------
The information contained in this communication (including any
attachments hereto) is confidential and is intended solely for the
personal and confidential use of the individual or entity to whom
it is addressed. The information may also constitute a legally
privileged confidential communication. If the reader of this
message is not the intended recipient or an agent responsible for
delivering it to the intended recipient, you are hereby notified
that you have received this communication in error and that any
review, dissemination, copying, or unauthorized use of this
information, or the taking of any action in reliance on the
contents of this information is strictly prohibited. If you have
received this communication in error, please notify us immediately
by e-mail, and delete the original message. Thank you

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to listserv@xxxxxxxxxxx with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html

.

Relevant Pages

  • Re: How many of you will be using your "tax rebate check" towards the pin hobby?
    ... OK, on soapbox: ... I fail to understand the mindset of people. ... the government says we "created 10 million new jobs" this ...
    (rec.games.pinball)
  • Re: OleDb delays commits and unlocks?
    ... As you state it now probably a simple merge can do the job (as long as you use the same tablenames and the same columnnames). ... Like you write it, it seems to me that you want to use simple tools for difficult problems and more basic tools for difficult jobs, as I try to say it in an anology. ... > requires two rename operations. ... > except on one of my test files. ...
    (microsoft.public.dotnet.framework.adonet)
  • Re: Name function in Access 2002 (XP)
    ... There are *myriads* of reasons that a rename can fail, ... the new file name is invalid; ... The VBA errors are very specific when there is a problem. ...
    (microsoft.public.access.formscoding)
  • Re: Let them fail...
    ... let them fail? ... the entire US economy which would in turn destabilize the world ... The jobs they create, and the trades they support all need them not to ... complete waste of time and money. ...
    (rec.gambling.poker)
  • Re: shell scrpt to rename
    ... How do I rename *.foo to *.bar? ... command line buffer), and will fail if any file names contain a ... In some linux distributions you may be able to use the rename ...
    (comp.unix.shell)