1. About Cheryl Watson’s Tuning Letter 2010 No. 2
2. Corrections to Tuning Letter 2010 No. 2
3. New Red Alerts
1. About Cheryl Watson’s Tuning Letter 2010 No. 2
The forty-seven page 2010 No. 2 Tuning Letter was emailed to paid subscribers on May 14th. You may visit our Web site at www.watsonwalker.com to obtain subscription information. The following is our Management Summary page from that issue, talking about some of the contents of this latest Tuning Letter:
System z Simplification
On October 4, 2006, IBM issued a press release describing a $100 million five-year effort to simplify the System z platform. This was announced with z/OS 1.8 and included facilities like the IBM Health Checker for z/OS (see page 27), the OMEGAMON z/OS Management Console (Tuning Letter 2006 No. 2, page 45), Hardware Configuration Manager (HCM) enhancements (Tuning Letter 2006 No. 4, page 30), IBM Configuration Assistant for z/OS Communications Server (page 5), application development simplification, and the z/OS Basic Skills Information Center (page 15). The result provides a z/OS platform that takes fewer people to manage, and ensures that z/OS continues to provide the best ROI in the industry.
Since that time, these facilities have been expanded and new facilities have been added. There is no charge for any of these new facilities, but you need to install them and configure them. Not enough installations have activated these, and it’s a shame because they will save sysprog and operator time after installation. In this issue, we describe the latest entries into this group of simplification tools – z/OSMF and PFA. z/OSMF (z/OS Management Facility) is a Web browser-based tool to let a system programmer install and maintain a sysplex in less time and using more automated tasks. In addition, this tool is free, and many of the functions can be run on a zIIP or zAAP if you have one available. PFA stands for the Predictive Failure Analysis, which is a tool that identifies soft failures or systems that are “dying, but not dead.” PFA is part of the z/OS base product and can reduce the risk of system failures.
Education
As z/OS continues to be the heart of most computer installations, and long-term system programmers and performance analysts are looking at retirement, education of new staff becomes more important than in the past. We know you have budget limitations for training, but that shouldn’t stop all training. The main thing that an employer has to provide is employee time to actually participate in the training. This is VERY, VERY, important, and we can’t stress it enough. Too many companies throw away their training budget because they don’t give an employee enough time to take advantage of the training they have available. According to Angelo Corridori of IDCP (see page 13) “Lack of time is the #1 reason that students don’t complete their courses. Employers are wasting their scarce training dollars if they agree to pay for a student’s classes and then load them up with work to the point they have no time to devote to the classes!”
In this Tuning Letter, we provide a list of training options that do not require travel expenses. Some are free, and none are unreasonable. In order to retain the many benefits of the mainframe, you need to start training the staff that will continue the maintenance of your System z software and hardware. In our article, you’ll learn about how one company is attacking the problem, and you’ll find some very low cost opportunities for training.
I’d like to point out one specific item from the education article relating to SHARE membership. Membership is a one-time fee of $500, which provides access to all proceedings and a reduction in conference fees. But in most companies, people aren’t aware that this is a company-wide membership. If you send one system programmer to a conference, or simply get a membership, do you publicize the information to other employees in the computer department? Most organizations don’t, inadvertently withholding information that could be invaluable to their DB2 DBA, the CICS administrators, the application programmers, and the operations support staff. The training available for this one-time fee is remarkable. And even fully-trained staff need SHARE to keep up with new releases, new problems, and so forth.
Elsewhere in this Issue
You’ll find many other useful items throughout this newsletter: User experiences on ALLOCxx, DB2 buffer pools, zFS, BCPii, HiperDispatch, and SMP/E • Health Checker and PFA (Predictive Failure Analysis). • Many New Function APARs, and several APARs about invalid SMF data.
2. Corrections to Tuning Letter 2010 No. 2
In our article on z/OSMF, we listed APAR PK97274 on page 11 and indicated that it applied to both z/OSMF and WAS OEM. In fact, the APAR for WAS OEM is PM02335, which we did not mention. This will be corrected for the DVD. As we mentioned in the Tuning Letter, it is extremely important that you apply PTFs for these APARs before attempting z/OSMF.
IBM APAR PK97274 (z/OSMF 1.11, 29Dec2009) – z/OSMF 1.11 Service Update 2.
IBM APAR PM02335 (WAS OEM 1.11, 17Jan2010) – Ship Upgrade for WebSphere Application Server OEM for z/OS V7.0.
IBM Manual – GA32-0631-01 – IBM WebSphere Application Server OEM Edition for z/OS Configuration Guide (the link is applicable to the manual which assumes you’ve applied APAR PM02335) (2010Jan)
3. Two Red Alerts
On April 30th and May 7th, IBM issued new Red Alerts. You can sign up to automatically receive Red Alerts at http://www14.software.ibm.com/webapp/set2/sas/f/redAlerts/home.html
2010.04.30 – z/OS with PTFs for OA30513 can experience data loss due to deletion of logstream datasets that are not eligible for deletion
With OA30513 applied, Logger dataset delete processing does not properly maintain the lowest valid point in the logstream leading to data loss. The following describes the circumstances when the problem can occur:
- PTFs for OA30513 are applied
- A logstream has allocated many offload datasets and is using more than one dataset directory or dataset directory extent
- This logstream is defined with AUTODELETE(YES) and a non-zero retention period (RETPD)
OR
This logstream is defined with AUTODELETE(NO), and the logstream’s data is being trimmed
When these conditions are met, there is a code defect where Logger can delete more datasets than it should. Applications attempting to browse the logstream can get back a gap condition or an error indicating the requested data cannot be found.
There are no Logger messages or displays that will externalize the loss of data. Please see APAR OA32737 for further details.
** Caution – OA30513 was marked as HIPER and closed 6Nov2009, so is probably in your latest RSU stream.
2010.05.07– Possible data loss for z/OS 1.11 users of GDG’s with GDS_RECLAIM option set to YES (default) in the SMS Parmlib member, IGDSMSxx
Only GDG users that have GDSs in DEFERRED ROLL IN STATUS and receive message IGD17356I GDG RECLAIM REQUEST WAS SUCCESSFULLY PROCESSED FOR DATASET in the joblog during job execution are affected. During RECLAIM processing, output intended to be directed to the data set is not written but the job receives a CC0.
Please see APAR OA32968 for additional information.
Stay Tuned!