Service levels, RSUs and change control for maintenance
Sites do not apply fixes one by one at random. They pick a tested service level such as an RSU, apply it to a cloned set of target libraries, test it, and roll it out system by system under change control. Emergency fixes follow a faster but still controlled path.
Target and distribution libraries
SMP/E manages two kinds of library for every product:
Systems do not IPL directly from the libraries being maintained. Most sites keep at least two sets of target volumes, often called system residence (SYSRES) sets: one running, one being prepared. Maintenance is applied to the spare set, which has its own target zone, and the set is then rolled out. The previous set is kept so a system can be IPLed back if the new level misbehaves.
Picking a service level: RSU and PUT
IBM ships fixes continuously, but also groups them into tested levels. Recommended Service Upgrade (RSU) levels identify PTFs that have been through additional cross-product testing in IBM's Consolidated Service Test (CST) environment. RSUs are identified by year and month, such as RSU2609, and IBM marks qualifying PTFs with SOURCEID values so you can select them. Older terms you will see are PUT levels, monthly groupings of PTFs.
SET BOUNDARY(ZOSTGT2).APPLY SOURCEID(RSU*)GROUPEXTENDBYPASS(HOLDSYSTEM)CHECK.Exact SOURCEID values and recommended practice vary by product and over time. Read the product's current preventive service guidance before relying on a pattern.
Preventive and corrective service
| Type | Purpose | Typical cadence |
|---|---|---|
| Preventive | Stay on a recent, tested level so known problems never hit you | Planned, often once or twice a year per product, at an RSU level |
| Corrective | Fix a problem you are actually seeing | As needed, sometimes as an emergency change |
| Security / HIPER | Install fixes flagged as security, high impact or pervasive | Prioritised; often outside the normal cycle |
IBM marks high-impact PTFs as HIPER and uses fix categories (FIXCATs) for security, hardware support and coexistence. REPORT ERRSYSMODS lists installed SYSMODs that now have ERROR holds against them, which is a useful regular check.
Change control around maintenance
- Back up first: target and distribution libraries, the CSI zones and SMPPTS, so the whole SMP/E environment can be recreated.
- Review: attach the APPLY CHECK reports, HOLDDATA review and the list of SYSTEM hold actions to the change record.
- Test: IPL a test system on the new level and run regression tests, including vendor products that sit on top of z/OS.
- Roll out gradually: sandbox, development, then production, one system at a time where a sysplex allows it.
- Back-out plan: name the previous SYSRES set and the steps to IPL back to it. See back-out plan.
- Record: note which service level each system runs, so support teams can answer 'is this fix on?' quickly.
Where SMP/E sits today
SMP/E is still how z/OS and most IBM Z software is installed and serviced. z/OSMF Software Management adds a web interface on top of SMP/E data for deploying software instances, reporting on installed service and finding missing fixes. Many products are now delivered as z/OSMF portable software instances as well as traditional SMP/E packages. Underneath, the CSI, zones and SYSMODs are the same, so the skills in this course still apply.
Common mistakes
The running system's libraries change underneath it and there is no clean way back. Always apply to a clone with its own target zone.
Isolated PTFs skip the cross-product testing of an RSU level. Use a tested service level for preventive maintenance and keep one-off fixes for corrective work.
RESTORE can be slow and has dependencies. The practical back-out is IPLing the previous SYSRES set, which must be kept and named in the change.
What you will see at work
- Maintenance changes are planned months ahead and often tied to a z/OS or subsystem upgrade project.
- Security teams request evidence that flagged security fixes are installed; SMP/E reports or z/OSMF provide it.
- When a vendor asks 'what service level are you on?', the answer comes from SMP/E records, not memory.
Key terms
Check your understanding.
Take this lesson's quiz and save your progress. Free.