name: ibm-mainframe
description: Use when working with IBM z/OS mainframe systems — JCL (Job Control Language) syntax and patterns, dataset types (sequential/PDS/PDSE/VSAM), IDCAMS utility, DFSORT/ICETOOL, ISPF navigation, TSO/REXX scripting, system utilities (IEBCOPY/IEBGENER/IEFBR14), RACF security basics, GDG (Generation Data Groups), SMS (Storage Management Subsystem), and z/OS fundamentals. Part of the ibm-mainframe skill family.
IBM z/OS Mainframe Fundamentals
For enterprise Python connectors to mainframe systems (Zowe SDK, ibm_db, EBCDIC handling) see: python-enterprise-connectors.
Always specify DISP on every DD statement — omitting DISP defaults to (NEW,DELETE) which loses your data. Explicitly code DISP=(NEW,CATLG,DELETE) for new datasets or DISP=SHR for existing ones.
Never submit a job that creates datasets without verifying the SPACE allocation. Under-allocation causes S B37/D37/E37 abends; over-allocation wastes DASD. Always estimate record count, record length, and blocking factor before coding SPACE.
Always use COND or IF/THEN/ELSE to control step execution — unconditional execution after failures causes data corruption. A failed step that writes partial data followed by an unconditional step that reads it will propagate bad data downstream.
Never delete a VSAM cluster without checking SHAREOPTIONS and verifying no other jobs have it open. Deleting an in-use VSAM cluster causes S213 abends in other jobs and potential data loss. Use LISTCAT to check the cluster status first.
Reference Files
Detailed code examples, patterns, and configuration are in the reference files below. Read the relevant file when working on that area.
| File |
Covers |
| datasets-idcams-dfsort.md |
dataset types (sequential, PDS, PDSE, VSAM), IDCAMS utility, and DFSORT/ICETOOL |
| ispf-rexx-racf-patterns.md |
ISPF navigation, TSO/REXX scripting, system utilities, RACF security, SMS storage management, and practical JCL patterns |
| zos-jcl.md |
z/OS fundamentals, JCL syntax and patterns (JOB, EXEC, DD statements, procedures, conditional execution, GDGs) |
Anti-Patterns
| Anti-Pattern |
Why It Fails |
Correct Approach |
| Submitting JCL without testing with TYPRUN=SCAN |
Syntax errors consume job queue slots and waste batch window time before failing |
Always validate JCL syntax with TYPRUN=SCAN or JCL checker before production submission |
| Using default SPACE allocation without growth estimates |
Under-allocated datasets get B37/D37 abends mid-job; over-allocated wastes DASD across the complex |
Calculate space from record length, volume, and growth; use RLSE to return unused space |
| Hardcoding dataset names without high-level qualifier substitution |
Jobs break when migrated between LPAR/environments; manual editing of DSNs is error-prone |
Use symbolic parameters and system symbols for HLQ; parameterize environment-specific DSNs |
| Not running IDCAMS VERIFY after abnormal job termination |
VSAM datasets left in an inconsistent state; subsequent jobs get locked-out or corrupt data |
Always run IDCAMS VERIFY on VSAM clusters after abend recovery before restarting dependent jobs |
| Ignoring return codes and condition codes in JCL |
COND parameter misuse causes steps to run (or skip) unexpectedly; silent data corruption |
Use IF/THEN/ELSE/ENDIF for clear condition handling; check RC after every step in multi-step jobs |
Related Skills
| Topic |
Skill |
| Python connectors to z/OS (Zowe SDK, ibm_db, EBCDIC) |
python-enterprise-connectors |
| DB2 for z/OS administration |
db2-mainframe |
By