This is from IBMs Year 2000 site. ill someone please translate and expand the ramifications of this story. Thanks Carl

Link to story:

http://www-4.ibm.com/software/year2000/alert/20000106164843.html

DFSMShsm: Non-SMS managed utility datasets may be incorrectly deleted; FREEVOL may cause loss of control data

SUBJECT: DFSMShsm: Non-SMS managed utility datasets may be incorrectly deleted; FREEVOL may cause loss of control data

. When utility datasets (a subset of temporary datasets) are defined as non-SMS managed, the following DFSMShsm functions: Primary Space Management, Interval Migration and Incremental Backup, may incorrectly delete these datasets. Please review APAR OW42637 to see the complete definition of utility datasets as well as additional information for determining your exposure.

In a multiple host environment, running the above indicated DFSMShsm functions on one system may delete a utility dataset in use on another system, causing that job to ABEND. This applies to any environment where there is more than one system image that can access the same DASD volume and is not limited to SYSPLEX environments.

2. The FREEVOL function is used to move the DFSMShsm data on a DASD volume to a new volume. FREEVOL processing deletes DFSMShsm control information for volume backups and dumps, rather than moving the data to another ML1 volume. RESOLUTION: A fix for APAR OW42637 is now available which resolves both problems.

The fix is available on FTP site

ftp://testcase.boulder.ibm.com/mvs/fromibm/. There are four different versions available, depending on system level:

-- Carl Jenkins (Somewherepress@aol.com), January 07, 2000

Answers

Obviously NOT a "Y2K problem".

It's just a simple "loss of control data" problem, that's all.

-- Ron Schwarz (rs@clubvb.com.delete.this), January 07, 2000.

From APAR OW42637:

..... ERROR DESCRIPTION: Two problems created by improper handling of data sets having a date field in their name are described in this APAR. . .....

..... Utility data sets are defined as non-SMS-managed data sets matching the following data set naming convention: . SYSnnnnn.Tnnnnnn.?Fccc SYSnnnnn.Tnnnnnn.?Vccc SYSnnnnn.Tnnnnnn.?Accc N---N.FORMAT.Dnnnnn.Tnnnnnncc N---N.SUBMIT.Dnnnnn.Tnnnnnncc N---N.MERGE.Dnnnnn.Tnnnnnncc N---N.RUN.Dnnnnn.Tnnnnnncc N---N.SUB.Dnnnnn.Tnnnnnncc N---N.R.Dnnnnn.Tnnnnnncc N---N.M.Dnnnnn.Tnnnnnncc N---N.SPRINT.Dnnnnn.Tnnnnnncc N---N.DOSPR.Dnnnnn.Tnnnnnncc *SYSUT where: - A string of ns is a string of numeric characters whose length is represented by the number of ns. This string usually consists of a time when preceded by a T or a date when preceded by a D. - A string of cs is an arbitrary length of arbitrary characters for the remaining portion of the data set name. - N---N represents the first qualifier in the data set name. - A string of ?s is a string of arbitrary characters whose length is represented by the number of ?s. .......

-- Earl Armbrust (Earl_Armbrust@compuserve.com), January 07, 2000.

This specific one is not a big problem.

There have been several Y2K problems reported post-roll in DFSMS. Similarly, CA7 has had three different Y2K problems reported post-roll.

What is revealing is that system software and system products have had newly discovered Y2K problems.

Note two things.

1) The pop-news has no reports on problems with DFSMShsm or CA7. You know there are problems. Joe six-pack doesn't.

2) These are data loss, unpredictable results, rerun the job situations. In some cases, these problems have corrupted production databases.

I've been sick with the flu since Monday and haven't been able to work on the systems as much as I've wanted. I have personally seen one of the CA7 problems.

I believe that DFSMS and CA7 are widely used. I'll write a technical evaluation on the real Y2K problem next week.

-- cory (kiyoinc@ibm.XOUT.net), January 07, 2000.

Aloha from Maui!!!, Cory, et al, Thanks for clarification on this issue!...I agree with you cory that the fact the media is mentioning nothing about the many y2k glitches is one of the most significant parts of this whole y2k story...It really points out the complicity of the mass media with their corporate sponsors/owners...Frankly, I think the talking heads should all have "paid advertisment" blinking below them every time they open their mouths! ... .anyway, off my soapbox...I'm going diggin'......

-- Carl Jenkins (Somewherepress@aol.com), January 07, 2000.

I heard Gerstner personally hired John Krepansky to fix this problem and lead IBM from the dark chaos in which they found themselves here recently.

Good luck in your new assignment John.

We're counting on you to help us all. Keep posting John, we need to know how YOU'RE doing on these issues.

-- Gordon (g_gecko_69@hotmail.com), January 07, 2000.

Hi all, sure wish I had a week on Maui coming up. I got the flu and haven't been able to do all the work my clients have asked for.

I should add for the non-mainframers, DFSMS is Data Facility / Storage Management System, just a bunch of software that manages storage in an MVS mainframe. CA7 is a powerful job scheduler manufactured by Computer Associates.

Both DFSMS and CA7 are very widely used and well accepted by the market. It's surprising that two important pieces of software have had multiple Y2K problems discovered after roll.

We'll see how this plays out. We could be lucky. I'm hoping for the best, as Ed Yourdon says.

-- cory (kiyoinc@ibm.XOUT.net), January 07, 2000.