Shell Scripting Fundamentals

Overview

“Write a small script” sounds scary until you see it is mostly saved commands with safety rails. Good Bash habits surface errors early instead of hiding them on production hosts.

Plain problem: A midnight cron script fails silently — no set -e, errors swallowed, exit code always 0. Monitoring thinks all is well. Good scripts fail loudly and leave evidence.

This page teaches the operations minimum: shebang, set -euo pipefail, arguments, exit codes, and readable output. Full Bash curriculum lives in the Shell Scripting track.

This is a Command Line tutorial in the REBASH Academy Linux for Cloud & DevOps Engineers series.

Prerequisites

• Ubuntu practice VM or WSL2
• Essential Linux Commands
• Text editor (nano, vim, or VS Code)

Learning Objectives

By the end of this tutorial, you will be able to:

• Explain what a shell script is in plain language
• Write a script with shebang and safe mode flags
• Use positional arguments ($1, $#) and exit codes
• Build a small host-check script that fails clearly
• Break a script on purpose and fix it
• Answer fresher interview questions on Bash scripting basics

Architecture

You write a text file → mark executable → shell reads lines top to bottom → each command returns an exit code → caller (cron/systemd/human) sees success or failure.

Theory

The problem (before any jargon)

Team script:

grep ERROR /var/log/app.log
echo "check done"

Log missing → grep fails → without set -e script continues → “check done” anyway → false green status.

What is a shell script? (simple words)

Analogy: A recipe card for the shell — step 1, step 2, same every time. Bash is the cook reading the card.

First line shebang picks the interpreter:

#!/usr/bin/env bash

Interview line: “I start ops scripts with set -euo pipefail so failures stop the script and unset variables error.”

Safe mode — set -euo pipefail

Flag	Meaning
-e	Exit on first command failure
-u	Error on unset variables
-o pipefail	Pipeline fails if any command fails

Arguments and exit codes

Terminal

echo "First arg: ${1:-none}"
exit 0   # success
exit 1   # generic failure

$? holds last exit code. Cron and systemd use it for success/failure.

Common pitfalls

• No shebang → wrong shell on cron
• Unquoted variables breaking on spaces
• Parsing ls output (use find or globs)
• Missing chmod +x

Hands-on Lab

Objective

Build host-check.sh with safe modes, arguments, intentional break, fix, and evidence under ~/rebash-linux/lab-shell.

Prerequisites

Item	Notes
Ubuntu VM	bash
Lab only	Script checks local disk

Lab environment

Terminal

mkdir -p ~/rebash-linux/lab-shell && cd ~/rebash-linux/lab-shell

Real-world scenario

Mentor: “Give me a script I can run from cron that checks root disk usage and exits non-zero if above a threshold — must not hide errors.”

Step-by-step tasks

Task 1 – host-check.sh (working version)

Create host-check.sh:

host-check.sh

#!/usr/bin/env bash
set -euo pipefail

THRESH="${1:-90}"
LOG="${HOME}/rebash-linux/lab-shell/host-check.log"
MOUNT="/"

usage="$(df -P "$MOUNT" | awk 'NR==2 {print $5}' | tr -d '%')"
{
  echo "=== $(date -Is) ==="
  echo "mount=$MOUNT usage_percent=$usage threshold=$THRESH"
} >> "$LOG"

if [[ "$usage" -ge "$THRESH" ]]; then
  echo "FAIL: disk usage ${usage}% >= ${THRESH}%" >&2
  exit 2
fi
echo "OK: disk usage ${usage}%"
exit 0

Terminal

cd ~/rebash-linux/lab-shell
chmod +x host-check.sh
./host-check.sh 90 | tee run-ok.txt
grep -q '^OK:' run-ok.txt
tail -3 host-check.log | tee log-ok-tail.txt
echo $? | tee exit-code-ok.txt

Expected output

OK: disk usage … printed; exit code 0; log appended.

Task 2 – Break (disable -e), observe silent failure

Create host-check-broken.sh:

host-check-broken.sh

#!/usr/bin/env bash
# intentionally missing set -e for lab break demo
THRESH="${1:-90}"
false
echo "This line should not run if -e were enabled"
exit 0

Terminal

cd ~/rebash-linux/lab-shell
chmod +x host-check-broken.sh
./host-check-broken.sh; echo "exit=$?" | tee broken-exit.txt
grep -q 'exit=0' broken-exit.txt
echo "break: script reported success after false command" | tee break-notes.txt

Expected output

Broken script prints misleading success line; exit=0 despite false — demonstrates why -e matters.

Task 3 – Fix threshold test and prove non-zero exit

Terminal

cd ~/rebash-linux/lab-shell
./host-check.sh 0; echo "exit=$?" | tee fail-threshold-exit.txt || true
grep -q 'exit=2' fail-threshold-exit.txt
./host-check.sh 90 | tee run-after-fix.txt
echo "lab-shell OK" | tee evidence.txt

Expected output

Threshold 0 forces FAIL exit code 2 (disk always >= 0%). Normal threshold returns OK again.

Validation steps

• host-check.sh uses shebang and set -euo pipefail
• Broken script demonstrates silent failure without -e
• Exit codes 0 vs 2 verified
• Log file receives timestamped entries

Common errors and fixes

Error	Cause	Fix
Permission denied	Not executable	chmod +x script.sh
bad interpreter	Windows CRLF line endings	dos2unix script.sh
Unbound variable	-u and missing arg	${1:-default}
Pipeline wrong status	Missing pipefail	set -o pipefail

Challenge exercise

Add a second check: fail if loadavg first field > 10 (use uptime or /proc/loadavg) — keep script under 40 lines.

Learning outcomes

• You wrote a production-shaped mini script
• You saw why safe modes matter
• You can discuss exit codes in interviews

Cleanup

Terminal

# Keep scripts and logs for revision

Validation

• Evidence under ~/rebash-linux/lab-shell
• Can explain set -euo pipefail in one sentence each
• Ready for environment variables tutorial next

Code Walkthrough

1. #!/usr/bin/env bash — portable shebang finding bash on PATH.
2. ${1:-90} — default threshold if no argument — avoids unset with -u.
3. df -P + awk — predictable parsing; avoid bare df locale surprises.
4. exit 2 — distinct code for disk threshold vs generic 1.
5. Broken script without -e — intentional anti-pattern for learning.

Security Considerations

• Quote variables: "$LOG" prevents word splitting/injection.
• Do not run curl|bash; review scripts before cron as root.
• Restrict script write permissions — attackers replace your script.
• Avoid secrets in scripts; use env files with tight permissions.
• Validate arguments ([[ "$THRESH" =~ ^[0-9]+$ ]]) before use.

Common Mistakes

❌ No set -e on ops scripts.

✅ Failures cascade silently — always use safe modes unless you handle each error.

---

❌ Unquoted $variables.

✅ Filenames with spaces break scripts; quoting is mandatory.

---

❌ Ignoring exit codes in cron.

✅ Cron only emails on failure if exit non-zero — return meaningful codes.