Shell Scripting — Performance & Optimisation

# ── time — measure overall script runtime ─────────────────
time ./backup.sh
# real    0m42.183s  ← wall clock time (what matters)
# user    0m0.284s   ← CPU time in user space
# sys     0m0.091s   ← CPU time in kernel

# ── Measure individual sections ───────────────────────────
time_section() {
  local label="${1}"; shift
  local start; start=$(date +%s%N)
  "$@"
  local elapsed=$(( ($(date +%s%N) - start) / 1000000 ))
  printf "  [TIMER] %-30s %dms\n" "${label}" "${elapsed}" >&2
}

time_section "DB dump"    mysqldump myapp > /backups/dump.sql
time_section "Compress"   gzip /backups/dump.sql
time_section "Upload"     rsync /backups/ bkp-01:/archive/

# ── Simple profiling with PS4 ─────────────────────────────
PS4='+ $(date "+%s%N") ${BASH_SOURCE[0]}:${LINENO}: '
exec 3>&2 2>/tmp/trace.log
set -x
# ... your script ...
set +x
exec 2>&3 3>&-
echo "Trace written to /tmp/trace.log"

# ── Every $() forks a subshell — expensive in loops ───────
# SLOW — forks cat + wc for every iteration
for f in /var/log/*.log; do
  lines=$(cat "${f}" | wc -l)   # 2 forks per iteration
  echo "${f}: ${lines} lines"
done

# FAST — wc reads files directly, one fork total
wc -l /var/log/*.log

# ── Useless use of cat ────────────────────────────────────
# SLOW
cat file.txt | grep "ERROR"      # forks cat unnecessarily
cat file.txt | wc -l              # same

# FAST — redirect directly
grep "ERROR" file.txt
wc -l < file.txt

# ── Use built-ins over external commands ──────────────────
# SLOW — forks an external process
n=$(expr $n + 1)
upper=$(echo "${str}" | tr '[:lower:]' '[:upper:]')

# FAST — bash built-in, no fork
(( n++ ))
upper="${str^^}"   # bash 4+ built-in uppercase

# ── More built-in alternatives ────────────────────────────
# Instead of: $(dirname "$path")  → use: "${path%/*}"
# Instead of: $(basename "$path") → use: "${path##*/}"
# Instead of: $(echo "$str" | sed 's/foo/bar/') → use: "${str/foo/bar}"
# Instead of: $(echo "$str" | grep -c 'pat') → use regex in [[ ]]

# ── xargs -P — parallel execution ─────────────────────────
# Compress 100 log files in parallel (8 at a time)
find /var/log -name "*.log" -mtime +7 | xargs -P 8 -I{} gzip {}

# Process images in parallel
find /photos -name "*.jpg" | \
  xargs -P $(nproc) -I{} convert {} -resize 800x600 {}_thumb.jpg

# ── Parallel with background jobs ─────────────────────────
MAX_PARALLEL=4
running=0

for server in "${servers[@]}"; do
  process_server "${server}" &

  (( ++running >= MAX_PARALLEL )) && {
    wait -n 2>/dev/null || wait   # wait for any one to finish
    (( running-- ))
  }
done
wait   # wait for all remaining

# ── Batch network checks in parallel ─────────────────────
check_host() {
  ping -c1 -W2 "${1}" &>/dev/null \
    && echo "UP:   ${1}" \
    || echo "DOWN: ${1}"
}
export -f check_host

printf "%s\n" web-{01..10} | xargs -P 10 -I{} bash -c 'check_host "$@"' _ {}

# ── Benchmark before/after parallelism ───────────────────
echo "Sequential:"
time { for i in {1..10}; do sleep 1; done; }
# real: ~10s

echo "Parallel:"
time { for i in {1..10}; do sleep 1 &; done; wait; }
# real: ~1s