Running and fixing a batch night
An operator's night is about one thing: getting the right jobs finished correctly before the business deadline. That means watching the critical path, fixing failures in the right order, rerunning safely, and telling people what is happening.
Before the night starts
- Check for known issues and changes going in tonight.
- Look at the plan: anything unusual, such as month end or a holiday calendar?
- Know your deadlines: which outputs must be ready, and by when (the SLA).
When a job fails
- Look at the job's output: return code, abend code and the messages around the failure.
- Decide if it is on the critical path. If so, it comes first.
- Find the cause before acting: bad data, missing file, space, security, or a code problem.
- Choose the action: fix and rerun, restart from a step, hold dependants, or escalate to the application on-call.
- Record what you did and why in the ticket.
Rerun or restart?
| Situation | Usual action |
|---|---|
| Failed in the first step, nothing written yet | Rerun from the top |
| Later step failed, earlier outputs are kept | Restart from the failing step |
| Job appends to files or updates a database | Check what was already written; a blind rerun can duplicate it |
| Data cannot be fixed tonight | Hold dependants, escalate, and agree a plan with the business |
Communicate early
If an SLA is at risk, say so before it is missed. A short update — what failed, what you are doing, when the next update comes — is worth more than silence followed by a perfect fix.
Hand over
At the end of the shift, list what failed, what was rerun or forced, what is still on hold and what needs follow-up. Good handovers stop the same failure surprising the next team.
Common mistakes
Most reruns of S0C7, space or security failures fail again. Understand the cause first.
Fix what is on the critical path first, even if it failed later.
A job left on hold with no note can block the next night. Record every hold and who owns it.
What you will see at work
- Most sites have a run book for critical jobs that says what to do if they fail.
- You will page application on-call staff when a fix needs a code or data change.
- Monthly and yearly processing nights are the busiest; extra staff and checks are normal.
Key terms
Check your understanding.
Take this lesson's quiz and save your progress. Free.