◆ PROTOTYPE / CLICKABLE MOCKUP — not published, not indexed. Use-case landing page per the SEO & Website Action Plan.
Home › Use Cases › Production Troubleshooting

Troubleshoot a Nonstop Production Issue in Minutes, Not Hours

See the programs, files, and call paths behind a failing transaction, then ask Navigator what the code is supposed to do.

The problem

It is 2am. A transaction is failing in production, and the program at fault has not been touched in years. The person who wrote it retired. The comments, if there are any, do not explain why it is structured this way. Your team is reverse-engineering decades-old TAL or COBOL under pressure, with no map of what else the program touches.

AtlasHow Atlas helps

Point Atlas at the failing program and it shows you, from the source:

NavigatorHow Navigator helps

Once you know what is involved, ask Navigator why. Instead of reading raw TAL or COBOL line by line under deadline pressure, ask what a paragraph is supposed to do, what rule governs a field, or how an error path works. Answers are generated from your analyzed source code.

The workflow

  1. Identify the failing program.
  2. Pull its Atlas impact map: callers, callees, file access, and risk flags.
  3. Ask Navigator what the relevant code path is supposed to do.
  4. Confirm root cause against actual business logic.
  5. Make the fix with Atlas showing what else it may touch.

Example scenario

Illustrative A typical incident, not a customer metric.

  1. 02:14

    Alert Settlement batch abends

    A nightly settlement batch on an HPE Nonstop (Tandem) system stops partway through its run.

  2. 02:16

    Atlas Map the failing program

    Atlas surfaces the failing COBOL program's callers, and a shared Enscribe file that another program also writes to.

  3. 02:21

    Navigator Ask what the rule is

    Navigator explains the validation on the amount field: the value should never exceed a defined ceiling.

  4. 02:26

    Atlas Trace the shared file back

    Following the shared Enscribe file to its other writer shows the second program was writing an out-of-range value.

  5. 02:30

    Root cause Found before any code changed

    The real cause sits in a different program than the one that abended — identified without touching either.

FAQ

Does TIC require software to be installed on the Nonstop system?
No. TIC performs read-only analysis of your source code off-platform. Nothing is installed or run on your Nonstop (Tandem) system.
Can we analyze one program before a full deployment?
Yes. You can start with a single failing program or a small representative set — no full rollout required.
How is Atlas different from manual impact analysis?
Manual cross-referencing takes days and still misses undocumented callers and shared files. Atlas maps callers, callees, file access, fields, and copybooks from the source automatically, and produces the same result every time.
How long does analysis take?
A representative set of 3–5 programs can be analyzed quickly for a proof-of-value; full-codebase runs scale from there.