# Demonstration: Every Step, In Order, On a Real Failure

## OBSERVE

A startup script fails: `/srv/demo/start.sh` exits with an error. That's the specific, concrete
symptom — not "the service is down."

## GATHER EVIDENCE

```
$ /srv/demo/start.sh
FATAL: config file not found: /srv/demo/config/settings.conf
```

The script's own error message IS the evidence — no guessing required.

## FORM HYPOTHESIS

The config file at that exact path doesn't exist. That's the specific, testable hypothesis —
not "something's misconfigured."

## TEST

```
$ ls /srv/demo/config/
ls: cannot access '/srv/demo/config/': No such file or directory
```

Confirmed — the directory itself doesn't exist yet, not just the file.

## REPAIR

```
$ mkdir -p /srv/demo/config
$ echo "mode=production" > /srv/demo/config/settings.conf
```

The minimal fix for the CONFIRMED cause — not a broader "reinstall everything" overcorrection.

## VALIDATE

```
$ /srv/demo/start.sh
Started with config: mode=production
```

The original symptom is gone, confirmed with real output, not assumption.

## PREVENT RECURRENCE

A real prevention step here: the script should probably check for and create its config
directory on first run, or the deployment process should ensure the config exists before the
service ever starts — naming this explicitly is what separates a real fix from a lucky one.

## Instructor Notes

**Demo goal:** Show all seven steps applied in strict order to one concrete, realistic failure
— the exact structure this module's Lab requires students to repeat themselves.

**Approximate timing:** 10 minutes live.

**Common mistake to watch for:** skipping straight to REPAIR (creating the file) without first
articulating the hypothesis and testing it — even when the fix seems obvious, naming the
hypothesis explicitly is the habit this module is building.