Background job failures
Background job failures mean expected batch work (month-end programs, interfaces, extracts) did not complete. This page covers diagnosing cancelled or failed jobs.
Monitor SM37 for cancellations (a failed nightly job hurts the business next morning), ensure enough background work processes, and build job dependencies (chains/events) so steps run in the right order.
- A scheduled job is cancelled or did not produce its expected result.
- In SM37, find the job and open its job log, it records why the job ended (and often points to a dump).
- Program error/dump (see ST22 from the job log).
- Watch out: Guessing instead of reading the log/dump.
Symptom
A scheduled job is cancelled or did not produce its expected result.
Start here: read the evidence
In SM37, find the job and open its job log, it records why the job ended (and often points to a dump). If it dumped, ST22 has the detail; if it failed on data/config, the job log message explains.
Likely causes, in order
- Program error/dump (see ST22 from the job log).
- Data/config issue, the program hit a business error.
- Resource, too few background work processes (job delayed) or memory.
- Dependency, a prerequisite job/step did not run.
How to fix it
- Open the job log in SM37; follow any dump to ST22.
- Fix the underlying program, data or config issue.
- Address resources (background work processes, memory) if relevant.
- Reschedule/repeat the job and confirm completion.
Preventing recurrence
Monitor SM37 for cancellations (a failed nightly job hurts the business next morning), ensure enough background work processes, and build job dependencies (chains/events) so steps run in the right order.
Common pitfalls
- Guessing instead of reading the log/dump.
- Fixing the symptom, not the root cause.
- Changing several things at once.