Cron Jobs in Linux Explained — Scheduling, Syntax and Real-World Patterns

# ─────────────────────────────────────────────────
# MANAGING YOUR CRONTAB — the essential commands
# ─────────────────────────────────────────────────

# Open your crontab in the default editor (usually nano or vi)
# This is the ONLY safe way to edit your crontab
crontab -e

# List all your current cron jobs — great for auditing
crontab -l

# Remove ALL your cron jobs (dangerous — no confirmation prompt!)
crontab -r

# Edit the crontab for a specific user (must run as root)
crontab -u deploy_user -e

# ─────────────────────────────────────────────────
# ANATOMY OF A CRON EXPRESSION
# ─────────────────────────────────────────────────
# Each job line follows this exact structure:
#
# ┌──────────── minute        (0 - 59)
# │  ┌─────────── hour          (0 - 23)
# │  │  ┌──────────── day of month  (1 - 31)
# │  │  │  ┌─────────── month         (1 - 12)
# │  │  │  │  ┌──────────── day of week   (0 - 7, both 0 and 7 = Sunday)
# │  │  │  │  │
# *  *  *  *  *  command_to_run

# ─────────────────────────────────────────────────
# REAL EXAMPLES with plain-English explanations
# ─────────────────────────────────────────────────

# Run database backup every day at 2:30am
30 2 * * * /opt/scripts/backup_postgres.sh

# Clear the application temp directory every hour
0 * * * * rm -rf /var/app/tmp/*

# Send a weekly analytics report every Monday at 9am
0 9 * * 1 /opt/scripts/send_weekly_report.py

# Run a health check every 15 minutes
*/15 * * * * /opt/scripts/health_check.sh

# Archive logs on the 1st of every month at midnight
0 0 1 * * /opt/scripts/archive_logs.sh

# Run a job only on weekdays (Mon-Fri) at 6am
0 6 * * 1-5 /opt/scripts/sync_business_data.sh

#!/bin/bash
# ─────────────────────────────────────────────────────────────────
# backup_postgres.sh
# Production-ready cron job: PostgreSQL backup with logging + locking
# Scheduled via crontab to run daily at 2:30am:
#   30 2 * * * /bin/bash /opt/scripts/backup_postgres.sh
# ─────────────────────────────────────────────────────────────────

# ── 1. EXPLICIT ENVIRONMENT ──────────────────────────────────────
# Cron's PATH is minimal (/usr/bin:/bin). Set it explicitly so
# commands like pg_dump, aws, gzip are all found correctly.
export PATH="/usr/local/bin:/usr/bin:/bin"

# Load app-specific environment variables from a secure file
# This file contains DB_USER, DB_NAME, S3_BUCKET etc.
# Permissions on this file should be 600 (owner read/write only)
source /etc/app/backup.env

# ── 2. LOGGING SETUP ─────────────────────────────────────────────
LOG_DIR="/var/log/app_backups"
LOG_FILE="${LOG_DIR}/backup_$(date +%Y-%m-%d).log"
TIMESTAMP=$(date '+%Y-%m-%d %H:%M:%S')

# Create log directory if it doesn't exist
mkdir -p "$LOG_DIR"

# Redirect ALL output (stdout + stderr) to the log file
# The tee command also prints to the terminal when run manually
exec > >(tee -a "$LOG_FILE") 2>&1

echo "[$TIMESTAMP] ─── Backup job started ───"

# ── 3. JOB LOCKING — prevent overlapping runs ────────────────────
LOCK_FILE="/tmp/backup_postgres.lock"

# flock acquires an exclusive lock on the lock file.
# -n means "non-blocking" — if the lock is already held, exit immediately.
# 9 is the file descriptor we're using for the lock.
exec 9>"$LOCK_FILE"
if ! flock -n 9; then
    echo "[$TIMESTAMP] Another backup is already running. Exiting."
    exit 1
fi

# ── 4. THE ACTUAL WORK ───────────────────────────────────────────
BACKUP_FILENAME="${DB_NAME}_$(date +%Y%m%d_%H%M%S).sql.gz"
BACKUP_PATH="/tmp/${BACKUP_FILENAME}"

echo "[$TIMESTAMP] Dumping database: $DB_NAME"

# pg_dump connects to PostgreSQL and streams SQL to gzip for compression
# PGPASSWORD is read from the sourced env file above
pg_dump -U "$DB_USER" -h "$DB_HOST" "$DB_NAME" | gzip > "$BACKUP_PATH"

# Check if the dump succeeded — never assume it worked
if [ $? -ne 0 ]; then
    echo "[$TIMESTAMP] ERROR: pg_dump failed! Aborting upload."
    exit 1
fi

echo "[$TIMESTAMP] Dump complete. Uploading to S3: s3://${S3_BUCKET}/"

# Upload to S3 — credentials come from the sourced env file
aws s3 cp "$BACKUP_PATH" "s3://${S3_BUCKET}/postgres-backups/${BACKUP_FILENAME}"

if [ $? -eq 0 ]; then
    echo "[$TIMESTAMP] Upload successful. Cleaning up local file."
    rm -f "$BACKUP_PATH"
else
    echo "[$TIMESTAMP] ERROR: S3 upload failed! Local backup retained at $BACKUP_PATH"
    exit 1
fi

echo "[$TIMESTAMP] ─── Backup job completed successfully ───"

# ─────────────────────────────────────────────────────────────────
# CRON NICKNAMES — readable shorthand for common schedules
# ─────────────────────────────────────────────────────────────────

@reboot   /opt/scripts/start_queue_worker.sh   # Run ONCE at system startup
@hourly   /opt/scripts/refresh_cache.sh         # Same as: 0 * * * *
@daily    /opt/scripts/cleanup_sessions.sh      # Same as: 0 0 * * *
@weekly   /opt/scripts/generate_sitemap.sh      # Same as: 0 0 * * 0
@monthly  /opt/scripts/invoice_generator.sh     # Same as: 0 0 1 * *
@yearly   /opt/scripts/archive_old_records.sh   # Same as: 0 0 1 1 *

# ─────────────────────────────────────────────────────────────────
# SYSTEM CRONTAB (/etc/crontab) — has an extra USER field
# ─────────────────────────────────────────────────────────────────
# Format: minute  hour  day  month  weekday  USER  command

# Run log rotation as root every day at 6:25am
25 6 * * *   root    /usr/sbin/logrotate /etc/logrotate.conf

# Run the app's data sync as the 'deploy' user at midnight
0  0 * * *   deploy  /opt/app/bin/sync_data.sh

# ─────────────────────────────────────────────────────────────────
# RUNNING A CRON JOB MANUALLY FOR TESTING
# ─────────────────────────────────────────────────────────────────
# The most reliable way to test a cron job is to simulate
# cron's environment: no shell customizations, minimal PATH.

# Run your script exactly as cron would — bare environment
env -i HOME=/root LOGNAME=root PATH=/usr/bin:/bin \
    /bin/bash /opt/scripts/backup_postgres.sh

# ─────────────────────────────────────────────────────────────────
# CHECKING CRON LOGS — when a job runs but you don't see output
# ─────────────────────────────────────────────────────────────────

# On systemd-based systems (Ubuntu 20+, RHEL 8+)
journalctl -u cron --since "1 hour ago"

# On older systems using syslog
grep CRON /var/log/syslog | tail -50

# On Red Hat / CentOS systems
grep CRON /var/log/cron | tail -50